LLM benchmark results, 2026-06-02

Metric Key

Quick reference for the columns in every table below.

Metric	Range	Direction	What it means
F1 (accuracy)	0 to 1	higher is better	Combined score of precision and recall against the human-verified ground-truth ad spans. F1 = 0 means the model found nothing right; F1 = 1 means it found every ad with the correct boundaries. Uses IoU >= 0.5 (predicted span must overlap truth span by at least half) to count a match.
Cost / episode	USD	lower is better	Average dollars per episode at the current pricing snapshot. Recomputed from token counts so all rows compare at the same prices regardless of when the call ran.
F1 / $	ratio	higher is better	F1 divided by cost-per-episode. Cheap accurate models score highest. Free-tier models are rank-listed separately because the ratio is undefined.
p50 / p95 latency	seconds	lower is better, with caveats	Median (p50) and tail (p95) wall-clock response time. Note: for models routed through OpenRouter (everything except claude-*), this includes OpenRouter’s queueing and upstream-provider latency, not just the model itself. Treat as a load/availability indicator, not a model-quality signal.
JSON compliance	0 to 1	higher is better	Fraction of responses that parsed as a clean JSON array matching the requested schema. 1.0 = always clean; lower = used object wrappers ({ads: [...]}), markdown fences, extra fields like sponsor, or required regex fallback to extract.
No-ad episode	PASS / FAIL	PASS desired	Negative-control test on ep-ai-cloud-essentials (which has no ads). PASS = zero predictions across all 15 windows. FAIL = the model false-positived on a non-ad segment, with the FP count shown.
F1 stdev	0 to 1	lower means more consistent	Standard deviation of F1 across the four ad-bearing episodes. High stdev = inconsistent across content types.
JSON mode	native / prompt-inject / mixed	–	How the model received its JSON-output instruction. native = provider accepted response_format=json_object for at least 95% of calls; prompt-inject = provider rejected it and the runner fell back to instructing JSON in the prompt for at least 95% of calls; mixed = neither path crossed the threshold (sample mostly comes from intermittent provider rejections). Reads from json_format_used in calls.jsonl. Useful when picking a model whose provider may not support native JSON mode – a strong JSON compliance score from a prompt-inject model carries different weight than the same score from a native model.

Glossary

• IoU (intersection over union): how much two time ranges overlap, expressed as (overlap) / (union). 0 means no overlap, 1 means identical ranges. We use IoU >= 0.5 as the threshold for a predicted ad to count as matching a truth ad.
• Trial: each (model, episode) pair runs 5 trials at temperature 0.0 to surface non-determinism. F1 numbers in tables are averaged across trials.
• Window: each episode is split into ~85-second sliding windows; the model judges each window independently. Per-window predictions are stitched together for episode-level scoring.
• Schema violations: number of times the response had at least one missing-required-field, wrong-type, or extra-key issue. Doesn’t tank F1, but signals brittleness.
• Extraction method: the route the parser took to recover the ad list. json_array_direct is the cleanest; method names with regex_* mean the JSON itself was malformed and we fell back to text matching.

TL;DR

Best Accuracy (F1 @ IoU >= 0.5)

All models ranked by F1 against human-verified ground truth. Cost includes free-tier models (shown at $0.00).

Rank	Model	F1	F1 stdev	Cost / episode	p50 latency	JSON compliance	JSON mode
1	qwen/qwen3.6-plus	0.693	0.040	$1.1119	39.9s	1.00	native
2	qwen/qwen3.5-plus-02-15	0.679	0.034	$1.2346	48.2s	1.00	native
3	qwen/qwen3.6-flash	0.660	0.087	$0.5461	13.0s	1.00	native
4	openai/gpt-5.5	0.647	0.061	$6.6806	6.4s	0.87	native
5	google/gemini-3.5-flash	0.625	0.022	$3.5200	5.2s	1.00	native
6	qwen/qwen3.5-27b	0.622	0.161	$3.2163	68.7s	0.85	native
7	openai/gpt-5.4	0.620	0.067	$2.5345	1.8s	0.81	native
8	google/gemini-2.5-pro	0.613	0.041	$3.8631	14.2s	0.97	native
9	openai/gpt-5.4-mini	0.584	0.112	$0.7644	1.2s	0.81	native
10	claude-opus-4-7	0.576	0.043	$7.8054	2.2s	1.00	prompt-inject
11	openai/o3	0.567	0.150	$3.0485	8.1s	0.92	native
12	x-ai/grok-4.3	0.506	0.105	$1.4987	3.9s	1.00	native
13	deepseek/deepseek-v4-flash	0.475	0.123	$0.1346	3.7s	0.81	native
14	google/gemma-4-31b-it	0.474	0.080	$0.1291	2.2s	0.85	native
15	deepseek/deepseek-r1	0.469	0.143	$1.1157	19.9s	0.97	native
16	claude-sonnet-4-6	0.446	0.042	$3.5376	1.4s	0.96	prompt-inject
17	qwen/qwen3-235b-a22b-2507	0.433	0.082	$0.0735	2.3s	0.79	native
18	openai/gpt-oss-120b	0.432	0.138	$0.0643	3.0s	0.70	native
19	deepseek/deepseek-r1-0528	0.420	0.142	$1.0181	16.5s	0.88	native
20	nvidia/llama-3.3-nemotron-super-49b-v1.5	0.411	0.135	$0.2159	24.2s	0.71	native
21	cohere/command-a	0.408	0.040	$2.6688	3.8s	0.71	native
22	google/gemini-3.1-flash-lite	0.402	0.039	$0.2762	0.8s	0.96	native
23	moonshotai/kimi-k2.6	0.400	0.164	$2.3369	35.3s	0.57	native
24	meta-llama/llama-4-maverick	0.398	0.016	$0.1508	1.1s	0.81	native
25	mistralai/mistral-medium-3.1	0.368	0.047	$0.4380	0.9s	1.00	native
26	claude-haiku-4-5-20251001	0.367	0.001	$1.2017	1.2s	0.60	prompt-inject
27	deepseek/deepseek-v4-pro	0.367	0.221	$0.6376	26.4s	0.87	native
28	mistralai/codestral-2508	0.314	0.049	$0.3241	0.7s	1.00	native
29	google/gemini-2.5-flash-lite	0.302	0.023	$0.1104	0.9s	0.97	native
30	deepseek/deepseek-v3.2	0.298	0.191	$0.2564	2.7s	0.88	native
31	google/gemini-2.5-flash	0.293	0.000	$0.3435	1.0s	1.00	native
32	openai/gpt-3.5-turbo	0.286	0.013	$0.5091	1.3s	0.71	native
33	meta-llama/llama-3.3-70b-instruct	0.274	0.081	$0.1008	1.5s	0.55	native
34	mistralai/mistral-large-2512	0.249	0.016	$0.5599	2.6s	1.00	native
35	qwen/qwen3-14b	0.248	0.166	$0.1261	20.9s	0.28	native
36	deepseek/deepseek-r1-distill-llama-70b	0.241	0.097	$0.7331	2.0s	0.74	native
37	meta-llama/llama-4-scout	0.235	0.109	$0.0807	0.8s	0.82	native
38	nvidia/nemotron-nano-9b-v2	0.221	0.114	$0.0813	12.0s	0.92	native
39	meta-llama/llama-3.1-8b-instruct	0.198	0.123	$0.0207	0.8s	0.85	native
40	cohere/command-r-plus-08-2024	0.102	0.035	$2.5787	1.0s	0.98	native
41	openai/o4-mini	0.075	0.125	$1.8999	6.9s	0.05	native
42	microsoft/phi-4	0.055	0.061	$0.0713	2.3s	0.86	native
43	qwen/qwen3-8b	0.005	0.011	$0.2578	59.4s	0.01	native
44	mistralai/mistral-7b-instruct-v0.1	0.000	0.000	$0.0285	7.1s	0.16	native

Best Value (F1 per dollar)

Paid-tier only. Free-tier models are excluded here because F1 / 0 is undefined; they are ranked separately under Best Free-Tier below.

Rank	Model	F1/$	F1	Cost / episode
1	meta-llama/llama-3.1-8b-instruct	9.54	0.198	$0.0207
2	openai/gpt-oss-120b	6.72	0.432	$0.0643
3	qwen/qwen3-235b-a22b-2507	5.90	0.433	$0.0735
4	google/gemma-4-31b-it	3.67	0.474	$0.1291
5	deepseek/deepseek-v4-flash	3.53	0.475	$0.1346
6	meta-llama/llama-4-scout	2.91	0.235	$0.0807
7	google/gemini-2.5-flash-lite	2.74	0.302	$0.1104
8	nvidia/nemotron-nano-9b-v2	2.72	0.221	$0.0813
9	meta-llama/llama-3.3-70b-instruct	2.71	0.274	$0.1008
10	meta-llama/llama-4-maverick	2.64	0.398	$0.1508
11	qwen/qwen3-14b	1.96	0.248	$0.1261
12	nvidia/llama-3.3-nemotron-super-49b-v1.5	1.90	0.411	$0.2159
13	google/gemini-3.1-flash-lite	1.46	0.402	$0.2762
14	qwen/qwen3.6-flash	1.21	0.660	$0.5461
15	deepseek/deepseek-v3.2	1.16	0.298	$0.2564
16	mistralai/codestral-2508	0.97	0.314	$0.3241
17	google/gemini-2.5-flash	0.85	0.293	$0.3435
18	mistralai/mistral-medium-3.1	0.84	0.368	$0.4380
19	microsoft/phi-4	0.77	0.055	$0.0713
20	openai/gpt-5.4-mini	0.76	0.584	$0.7644
21	qwen/qwen3.6-plus	0.62	0.693	$1.1119
22	deepseek/deepseek-v4-pro	0.57	0.367	$0.6376
23	openai/gpt-3.5-turbo	0.56	0.286	$0.5091
24	qwen/qwen3.5-plus-02-15	0.55	0.679	$1.2346
25	mistralai/mistral-large-2512	0.45	0.249	$0.5599
26	deepseek/deepseek-r1	0.42	0.469	$1.1157
27	deepseek/deepseek-r1-0528	0.41	0.420	$1.0181
28	x-ai/grok-4.3	0.34	0.506	$1.4987
29	deepseek/deepseek-r1-distill-llama-70b	0.33	0.241	$0.7331
30	claude-haiku-4-5-20251001	0.31	0.367	$1.2017
31	openai/gpt-5.4	0.24	0.620	$2.5345
32	qwen/qwen3.5-27b	0.19	0.622	$3.2163
33	openai/o3	0.19	0.567	$3.0485
34	google/gemini-3.5-flash	0.18	0.625	$3.5200
35	moonshotai/kimi-k2.6	0.17	0.400	$2.3369
36	google/gemini-2.5-pro	0.16	0.613	$3.8631
37	cohere/command-a	0.15	0.408	$2.6688
38	claude-sonnet-4-6	0.13	0.446	$3.5376
39	openai/gpt-5.5	0.10	0.647	$6.6806
40	claude-opus-4-7	0.07	0.576	$7.8054
41	cohere/command-r-plus-08-2024	0.04	0.102	$2.5787
42	openai/o4-mini	0.04	0.075	$1.8999
43	qwen/qwen3-8b	0.02	0.005	$0.2578
44	mistralai/mistral-7b-instruct-v0.1	0.00	0.000	$0.0285

Charts

Cost vs F1 (Pareto)

Each model is one colored point. Lower-left is unhelpful (expensive, inaccurate). Upper-left is the sweet spot (accurate, cheap). The legend below the chart shows each model’s color next to its F1 and cost-per-episode.

Source data: Best Accuracy, Best Value, Best Free-Tier

JSON schema compliance

Fraction of each model’s responses that parsed as a clean JSON array. 1.0 means every response came back exactly as requested; lower numbers mean the parser had to recover from markdown fences, object wrappers, or extra fields.

Source data: Per-Model Detail (JSON compliance field)

F1 by episode (heatmap)

F1 score for each (model, episode) pair. Greener is more accurate, redder is less. The no-ad episode is excluded. It has no F1 because it’s a PASS/FAIL negative control.

Source data: Quick Comparison, Per-Episode Detail

Confidence calibration (heatmap)

One row per model, one column per self-reported confidence bin. Cell text is the actual hit rate at that bin plus the sample size; cell color is the calibration error (actual minus bin midpoint). Red cells mean the model claimed high confidence but was usually wrong; green is well-calibrated; blue is underconfident. Empty cells mean the model never produced a prediction in that bin. Models are sorted from most overconfident at the top to most underconfident at the bottom.

Source data: Confidence calibration table

Latency percentiles

p50, p90, p99, and max per model on a log scale. The gap between p99 and max indicates how heavy the tail is. For OpenRouter-routed models, the tail also includes upstream provider load.

Source data: Latency tail table

Cross-model agreement (window distribution)

Histogram of how many models flagged at least one ad per (episode, window). The left side is windows nobody flagged (clear non-ad content), the right side is windows everyone flagged (clear sponsor reads). Bars in the middle are contested (some models said yes, some said no) and are candidates for ensemble voting or manual review. This view is anonymous (bars don’t show which models contributed); the per-model breakdown is in the next chart.

Source data: Cross-model agreement table

Per-model alignment with majority

Stacked horizontal bar per model. Green + blue segments are windows where the model voted with the majority (true positives + true negatives); orange is windows where it voted yes but most others voted no (likely false positive / hallucination); red is windows where it voted no but most others voted yes (likely missed real ad). Right-edge label is alignment rate. High alignment means the model tracks consensus; low alignment is either insight or noise depending on whether those broken-from-consensus calls were right.

Source data: Per-model alignment with consensus table

Precision vs Recall (with F1 isocurves)

Scatter of precision (y) vs recall (x) for each model. Dashed gray lines are F1 isocurves; points on the same dashed line have the same F1. Top-right is ideal (high precision AND high recall). Top-left is cautious (high precision, low recall). Bottom-right is greedy (high recall, low precision). Useful for picking a model whose error profile matches your tolerance: precision-leaning for environments where false positives are expensive, recall-leaning for completeness-first.

Source data: Precision, recall, and FP/FN breakdown table

Boundary accuracy (start + end MAE)

Stacked horizontal bars per model: blue is mean absolute error on the predicted ad START in seconds, orange is the same for END. Total error labeled at the right. Sorted by total ascending so the cleanest boundaries are at the top. Skewed bars (start much larger than end, or vice versa) mean the model systematically overshoots on one side. Relevant if you cut audio downstream.

Source data: Boundary accuracy table

Token efficiency vs F1

Scatter of output tokens per detected ad (x, log scale) vs F1 (y). Upper-left is the efficient zone: high accuracy with few output tokens. Right-side points are reasoning-heavy models that emit chain-of-thought alongside their JSON. The chart answers whether the extra tokens buy more F1 or just burn output budget. A model that lands far right at modest F1 is paying for reasoning that didn’t help.

Source data: Output token efficiency table

Trial variance (determinism check)

Horizontal bars of mean F1 stdev across episodes per model. All trials run at temperature 0.0 so well-behaved models cluster near zero. Bars are color-graded: green below 0.02 (effectively deterministic), yellow 0.02-0.05 (slight noise), red above 0.05 (single-trial F1 numbers from this model should be treated with suspicion). Dotted reference lines mark the 0.02 and 0.05 thresholds.

Source data: Trial variance (determinism check) table

Detection rate by ad length

Heatmap of model (row) vs ad-length bucket (column), cell = detection rate with sample size. Greener = caught more ads in that bucket; redder = missed more. Models are sorted by overall detection rate so the strongest are at the top. Empty (gray) cells mean that bucket had no truth ads for the corresponding model’s trials.

Source data: Detection rate by ad characteristic > By ad length table

Detection rate by ad position

Same shape as the ad-length heatmap, but columns are episode position (pre-roll / mid-roll / post-roll). A common pattern: pre-roll is easy because of clear show-intro transitions; post-roll is harder because models near the end of long episodes often produce shorter responses or run out of context to anchor on.

Source data: Detection rate by ad characteristic > By ad position table

Parser stress (extraction-method usage)

Heatmap of model (row) vs extraction-method (column), cell = number of responses parsed via that method. Columns are ordered by total usage. json_array_direct is the clean path; everything else is a recovery path the parser had to take because the model added markdown fences, wrapped the array in an object, or returned malformed JSON. Models near the top of the chart use the clean path most often. They are operationally easier to consume.

Source data: Parser stress test table

Failures and provider issues

3 call(s) failed out of 37620 total (0.01%). Failures are excluded from F1 / cost calculations, but they often surface real production-relevant gotchas worth knowing.

By category

Errors classified into coarse buckets so failure patterns are visible at a glance. A model showing up here doesn’t mean it’s broken. Some categories are provider-side (content moderation, rate limits) and tell you more about routing reliability than model quality.

Category	Calls	Affected models
Provider content moderation rejection	3	qwen/qwen3.5-plus-02-15

Per-model error count

Same errors grouped by model, with the failure rate as a fraction of that model’s total calls. Rates under 1% are usually one-off provider hiccups; rates above 5% suggest the model isn’t operationally viable for production with the current prompts and concurrency caps.

Model	Errors	of total
qwen/qwen3.5-plus-02-15	3	3/855 (0.4%)

Sample messages (first 3 per category)

First three raw error messages per category, so you can see what the provider actually returned without grepping calls.jsonl. Messages are truncated to ~240 characters; full text lives in results/raw/calls.jsonl.

Provider content moderation rejection (3)

• qwen/qwen3.5-plus-02-15 on ep-drink-champs-30c9a2d49f13 (trial 0, window 15): Error code: 400 - {‘error’: {‘message’: ‘Provider returned error’, ‘code’: 400, ‘metadata’: {‘raw’: ’{“error”:{“message”:“<400> InternalError.Algo.DataInspectionFailed: Input text data may contain inappropriate content.”,“type”:“data_inspec…
• qwen/qwen3.5-plus-02-15 on ep-drink-champs-30c9a2d49f13 (trial 1, window 15): Error code: 400 - {‘error’: {‘message’: ‘Provider returned error’, ‘code’: 400, ‘metadata’: {‘raw’: ’{“error”:{“message”:“<400> InternalError.Algo.DataInspectionFailed: Input text data may contain inappropriate content.”,“type”:“data_inspec…
• qwen/qwen3.5-plus-02-15 on ep-drink-champs-30c9a2d49f13 (trial 3, window 15): Error code: 400 - {‘error’: {‘message’: ‘Provider returned error’, ‘code’: 400, ‘metadata’: {‘raw’: ’{“error”:{“message”:“<400> InternalError.Algo.DataInspectionFailed: Input text data may contain inappropriate content.”,“type”:“data_inspec…

Why this section exists

If you’re picking a model for production, an aggregate compliance score doesn’t tell you when the provider will simply refuse to answer. A few cases that have shown up here:

• Content moderation rejections (Alibaba on Qwen, Google on Gemma, sometimes others): the provider’s classifier blocks the prompt before the model runs. For ad detection on real podcast transcripts, this can happen on episodes with adult content, profanity, or politically sensitive topics. Rate is small but non-zero; plan for it.
• Deprecated parameters: the Claude 4.x family rejects temperature. The benchmark memoizes this per-process and retries without, but it tells you which models you cannot pass legacy sampling controls to.
• Rate limits: tail-latency or 429s under load. Not a model-quality issue, but determines whether a given provider is operationally viable for your throughput.

Precision, recall, and FP/FN breakdown

F1 collapses two failure modes into one number. A precision-leaning model misses ads but rarely flags non-ads; a recall-leaning model catches everything at the cost of false positives. Production tradeoffs hinge on which one you can tolerate.

Column key

Column	Meaning	Range
TP (true positive)	Predicted an ad and a real ad existed at that span (IoU >= 0.5)	0 to total truth ads
FP (false positive)	Predicted an ad where no real ad existed	0 to total predictions
FN (false negative)	Missed a real ad entirely (no prediction matched it at IoU >= 0.5)	0 to total truth ads
Precision	TP / (TP + FP). Of the ads the model claimed, how many were real? Higher means fewer false positives.	0.000 to 1.000
Recall	TP / (TP + FN). Of the real ads, how many did the model find? Higher means fewer misses.	0.000 to 1.000

Reading the table: high precision + low recall means the model is cautious. It rarely flags something that isn’t an ad, but misses real ads. High recall + low precision means the opposite: catches everything but invents false positives. F1 is the harmonic mean of the two and rewards models that do both well.

Model	Precision	Recall	TP	FP	FN
qwen/qwen3.6-plus	0.640	0.823	205	164	50
qwen/qwen3.5-plus-02-15	0.621	0.823	183	151	45
qwen/qwen3.6-flash	0.569	0.836	205	209	50
openai/gpt-5.5	0.600	0.760	185	144	70
google/gemini-3.5-flash	0.557	0.775	186	193	69
qwen/qwen3.5-27b	0.629	0.680	162	145	93
openai/gpt-5.4	0.553	0.781	190	195	65
google/gemini-2.5-pro	0.523	0.822	205	229	50
openai/gpt-5.4-mini	0.523	0.756	180	236	75
claude-opus-4-7	0.526	0.705	161	180	94
openai/o3	0.715	0.516	137	45	118
x-ai/grok-4.3	0.410	0.744	178	417	77
deepseek/deepseek-v4-flash	0.376	0.749	177	419	78
google/gemma-4-31b-it	0.382	0.690	158	407	97
deepseek/deepseek-r1	0.378	0.697	167	407	88
claude-sonnet-4-6	0.380	0.633	161	435	94
qwen/qwen3-235b-a22b-2507	0.351	0.632	146	388	109
openai/gpt-oss-120b	0.364	0.638	147	433	108
deepseek/deepseek-r1-0528	0.370	0.678	157	665	98
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.411	0.488	110	260	145
cohere/command-a	0.339	0.631	134	376	121
google/gemini-3.1-flash-lite	0.293	0.709	152	658	103
moonshotai/kimi-k2.6	0.468	0.433	91	122	164
meta-llama/llama-4-maverick	0.346	0.540	137	273	118
mistralai/mistral-medium-3.1	0.327	0.518	123	523	132
claude-haiku-4-5-20251001	0.265	0.625	145	656	110
deepseek/deepseek-v4-pro	0.461	0.363	98	115	157
mistralai/codestral-2508	0.278	0.474	116	546	139
google/gemini-2.5-flash-lite	0.208	0.607	141	771	114
deepseek/deepseek-v3.2	0.307	0.332	83	259	172
google/gemini-2.5-flash	0.207	0.538	125	800	130
openai/gpt-3.5-turbo	0.237	0.458	103	518	152
meta-llama/llama-3.3-70b-instruct	0.273	0.329	87	240	168
mistralai/mistral-large-2512	0.168	0.566	128	1073	127
qwen/qwen3-14b	0.312	0.238	60	132	195
deepseek/deepseek-r1-distill-llama-70b	0.216	0.361	83	449	172
meta-llama/llama-4-scout	0.212	0.329	81	459	174
nvidia/nemotron-nano-9b-v2	0.208	0.282	65	490	190
meta-llama/llama-3.1-8b-instruct	0.209	0.252	55	960	200
cohere/command-r-plus-08-2024	0.128	0.102	34	94	221
openai/o4-mini	0.183	0.049	12	21	243
microsoft/phi-4	0.051	0.076	21	484	234
qwen/qwen3-8b	0.017	0.003	1	5	254
mistralai/mistral-7b-instruct-v0.1	0.000	0.000	0	0	255

Boundary accuracy

For ads that match the truth at IoU >= 0.5, how far off were the predicted start and end timestamps? Lower is better. A model can hit F1 cleanly while still being 20s off on every boundary. Bad for any pipeline that cuts the audio.

Model	Start MAE (s)	End MAE (s)
qwen/qwen3-8b	0.02	0.01
qwen/qwen3.5-plus-02-15	5.43	1.26
qwen/qwen3.6-flash	5.40	2.59
google/gemini-3.5-flash	4.13	3.89
google/gemini-2.5-pro	4.83	3.55
qwen/qwen3.6-plus	5.02	3.38
google/gemini-3.1-flash-lite	1.81	7.28
x-ai/grok-4.3	3.38	6.38
nvidia/llama-3.3-nemotron-super-49b-v1.5	4.99	4.98
qwen/qwen3.5-27b	5.43	4.55
claude-sonnet-4-6	1.98	8.26
openai/gpt-5.4	7.02	3.50
deepseek/deepseek-v3.2	3.85	6.72
claude-haiku-4-5-20251001	2.24	8.38
openai/gpt-5.5	7.16	3.61
deepseek/deepseek-r1	4.50	6.29
google/gemini-2.5-flash	4.22	6.99
deepseek/deepseek-r1-0528	4.93	6.51
deepseek/deepseek-v4-pro	6.31	5.57
moonshotai/kimi-k2.6	9.52	2.37
openai/o4-mini	4.41	7.99
mistralai/mistral-large-2512	5.18	7.28
cohere/command-a	4.82	7.65
openai/o3	8.74	3.94
mistralai/codestral-2508	2.47	11.68
openai/gpt-oss-120b	4.55	9.60
meta-llama/llama-4-scout	3.72	10.74
deepseek/deepseek-r1-distill-llama-70b	2.23	12.34
claude-opus-4-7	9.86	5.63
microsoft/phi-4	8.58	7.07
mistralai/mistral-medium-3.1	6.74	10.03
openai/gpt-5.4-mini	6.31	10.47
google/gemma-4-31b-it	9.27	7.75
qwen/qwen3-235b-a22b-2507	6.70	10.72
deepseek/deepseek-v4-flash	7.29	10.23
meta-llama/llama-4-maverick	4.22	13.74
qwen/qwen3-14b	6.00	12.42
meta-llama/llama-3.3-70b-instruct	3.73	14.69
cohere/command-r-plus-08-2024	6.29	12.19
google/gemini-2.5-flash-lite	6.38	12.92
openai/gpt-3.5-turbo	6.40	13.03
meta-llama/llama-3.1-8b-instruct	12.70	6.99
nvidia/nemotron-nano-9b-v2	15.54	4.44

Confidence calibration

Models include a self-reported confidence on each detected ad. A well-calibrated model should be right ~95% of the time when it claims 0.95 confidence. The table below bins each model’s predictions and shows the actual hit rate (fraction that were true positives at IoU >= 0.5). A bin near 1.0 is well-calibrated; a low number with a high count means the model is overconfident.

Model	0.00-0.70	0.70-0.90	0.90-0.95	0.95-0.99	0.99+	total
claude-haiku-4-5-20251001	–	0.00 (n=64)	0.12 (n=264)	0.24 (n=473)	–	801
claude-opus-4-7	0.00 (n=2)	0.00 (n=7)	0.26 (n=19)	0.49 (n=290)	0.65 (n=23)	341
claude-sonnet-4-6	–	0.04 (n=47)	0.23 (n=84)	0.27 (n=422)	0.56 (n=43)	596
cohere/command-a	0.00 (n=2)	0.00 (n=38)	–	0.28 (n=481)	0.00 (n=1)	522
cohere/command-r-plus-08-2024	–	–	0.00 (n=1)	0.07 (n=54)	0.41 (n=73)	128
deepseek/deepseek-r1	–	0.00 (n=4)	0.17 (n=12)	0.28 (n=435)	0.34 (n=129)	580
deepseek/deepseek-r1-0528	0.00 (n=2)	0.00 (n=19)	0.02 (n=59)	0.10 (n=587)	0.47 (n=204)	871
deepseek/deepseek-r1-distill-llama-70b	–	0.02 (n=190)	–	0.20 (n=396)	–	586
deepseek/deepseek-v3.2	0.00 (n=1)	0.00 (n=5)	0.00 (n=10)	0.10 (n=188)	0.44 (n=144)	348
deepseek/deepseek-v4-flash	0.00 (n=7)	0.09 (n=11)	0.08 (n=12)	0.27 (n=370)	0.38 (n=197)	597
deepseek/deepseek-v4-pro	0.00 (n=1)	0.00 (n=11)	0.00 (n=4)	0.49 (n=156)	0.54 (n=41)	213
google/gemini-2.5-flash	–	–	0.00 (n=60)	0.15 (n=698)	0.12 (n=167)	925
google/gemini-2.5-flash-lite	–	0.50 (n=2)	–	0.15 (n=905)	0.50 (n=10)	917
google/gemini-2.5-pro	–	0.00 (n=16)	0.00 (n=31)	0.34 (n=119)	0.60 (n=274)	440
google/gemini-3.1-flash-lite	–	0.00 (n=3)	0.00 (n=33)	0.10 (n=177)	0.22 (n=602)	815
google/gemini-3.5-flash	–	0.00 (n=3)	0.12 (n=25)	0.10 (n=48)	0.59 (n=303)	379
google/gemma-4-31b-it	–	0.10 (n=10)	0.09 (n=33)	0.18 (n=234)	0.38 (n=293)	570
meta-llama/llama-3.1-8b-instruct	–	0.00 (n=5)	0.00 (n=2)	0.05 (n=1008)	–	1015
meta-llama/llama-3.3-70b-instruct	–	0.00 (n=22)	0.23 (n=31)	0.12 (n=137)	0.46 (n=137)	327
meta-llama/llama-4-maverick	0.00 (n=1)	0.00 (n=60)	0.10 (n=52)	0.44 (n=300)	0.00 (n=2)	415
meta-llama/llama-4-scout	–	0.00 (n=6)	0.05 (n=20)	0.15 (n=489)	0.32 (n=25)	540
microsoft/phi-4	–	0.00 (n=24)	0.00 (n=18)	0.04 (n=521)	–	563
mistralai/codestral-2508	–	0.00 (n=1)	0.00 (n=4)	0.18 (n=656)	0.00 (n=1)	662
mistralai/mistral-large-2512	0.00 (n=2)	0.00 (n=33)	0.00 (n=61)	0.05 (n=542)	0.18 (n=563)	1201
mistralai/mistral-medium-3.1	–	0.00 (n=6)	0.00 (n=57)	0.20 (n=560)	0.43 (n=23)	646
moonshotai/kimi-k2.6	0.00 (n=35)	0.04 (n=28)	0.00 (n=2)	0.49 (n=93)	0.65 (n=68)	226
nvidia/llama-3.3-nemotron-super-49b-v1.5	–	0.00 (n=6)	0.06 (n=32)	0.32 (n=330)	0.50 (n=2)	370
nvidia/nemotron-nano-9b-v2	–	0.00 (n=6)	0.00 (n=28)	0.12 (n=503)	0.32 (n=19)	556
openai/gpt-3.5-turbo	–	–	0.00 (n=5)	0.19 (n=441)	0.08 (n=231)	677
openai/gpt-5.4	0.00 (n=19)	0.08 (n=37)	0.00 (n=25)	0.39 (n=79)	0.67 (n=233)	393
openai/gpt-5.4-mini	0.00 (n=15)	0.02 (n=49)	0.06 (n=35)	0.41 (n=95)	0.61 (n=226)	420
openai/gpt-5.5	0.00 (n=12)	0.17 (n=18)	0.38 (n=8)	0.52 (n=73)	0.64 (n=219)	330
openai/gpt-oss-120b	–	0.00 (n=6)	0.11 (n=19)	0.15 (n=267)	0.36 (n=293)	585
openai/o3	0.00 (n=1)	–	0.31 (n=16)	0.79 (n=156)	1.00 (n=9)	182
openai/o4-mini	–	0.00 (n=2)	0.00 (n=1)	0.41 (n=29)	0.00 (n=1)	33
qwen/qwen3-14b	–	–	0.00 (n=31)	0.37 (n=163)	–	194
qwen/qwen3-235b-a22b-2507	0.00 (n=12)	0.00 (n=5)	0.00 (n=8)	0.28 (n=471)	0.22 (n=64)	560
qwen/qwen3-8b	–	–	–	0.17 (n=6)	–	6
qwen/qwen3.5-27b	–	0.00 (n=9)	0.12 (n=8)	0.56 (n=288)	0.00 (n=2)	307
qwen/qwen3.5-plus-02-15	–	0.00 (n=13)	0.11 (n=9)	0.58 (n=295)	0.71 (n=17)	334
qwen/qwen3.6-flash	0.00 (n=1)	0.00 (n=6)	0.00 (n=9)	0.51 (n=394)	0.83 (n=6)	416
qwen/qwen3.6-plus	0.00 (n=1)	0.00 (n=9)	0.00 (n=4)	0.59 (n=347)	0.25 (n=8)	369
x-ai/grok-4.3	0.00 (n=1)	0.07 (n=15)	0.10 (n=51)	0.32 (n=497)	0.35 (n=31)	595

See report_assets/calibration.svg for the visual reliability diagram.

Latency tail

Median latency hides outliers. p99 and max are what determines queue depth and worst-case user wait. For OpenRouter-routed models the tail also reflects upstream provider load, not just model compute.

Model	p50	p90	p95	p99	max
mistralai/codestral-2508	0.73s	1.72s	2.24s	4.67s	6.36s
google/gemini-3.1-flash-lite	0.79s	1.27s	1.44s	1.96s	17.47s
meta-llama/llama-3.1-8b-instruct	0.79s	2.39s	4.04s	7.07s	76.80s
meta-llama/llama-4-scout	0.84s	3.23s	4.44s	6.25s	16.81s
mistralai/mistral-medium-3.1	0.91s	4.84s	6.18s	8.28s	11.71s
google/gemini-2.5-flash-lite	0.92s	1.92s	3.23s	6.53s	17.01s
cohere/command-r-plus-08-2024	0.95s	2.23s	3.45s	14.64s	62.06s
google/gemini-2.5-flash	0.99s	2.45s	3.59s	10.06s	16.44s
meta-llama/llama-4-maverick	1.06s	2.06s	2.38s	3.94s	50.80s
claude-haiku-4-5-20251001	1.17s	3.18s	4.06s	181.18s	186.76s
openai/gpt-5.4-mini	1.18s	1.66s	2.24s	3.13s	4.94s
openai/gpt-3.5-turbo	1.26s	1.77s	1.95s	2.58s	8.25s
claude-sonnet-4-6	1.44s	4.71s	6.04s	8.75s	185.04s
meta-llama/llama-3.3-70b-instruct	1.47s	3.22s	4.82s	14.03s	45.01s
openai/gpt-5.4	1.82s	2.58s	3.12s	4.81s	17.20s
deepseek/deepseek-r1-distill-llama-70b	1.98s	15.01s	51.21s	94.11s	136.13s
claude-opus-4-7	2.17s	3.56s	4.21s	6.25s	181.97s
google/gemma-4-31b-it	2.21s	14.13s	19.39s	66.10s	377.81s
qwen/qwen3-235b-a22b-2507	2.28s	6.26s	7.75s	11.89s	24.93s
microsoft/phi-4	2.29s	6.71s	11.55s	203.39s	229.43s
mistralai/mistral-large-2512	2.55s	5.40s	6.20s	8.68s	18.09s
deepseek/deepseek-v3.2	2.67s	5.26s	7.32s	12.54s	63.86s
openai/gpt-oss-120b	2.96s	20.95s	34.40s	384.58s	2880.77s
deepseek/deepseek-v4-flash	3.67s	19.68s	28.76s	47.61s	80.55s
cohere/command-a	3.78s	8.49s	11.46s	30.31s	65.77s
x-ai/grok-4.3	3.88s	9.63s	12.56s	18.90s	33.29s
google/gemini-3.5-flash	5.23s	9.10s	11.13s	15.80s	32.29s
openai/gpt-5.5	6.38s	18.56s	24.05s	37.19s	76.20s
openai/o4-mini	6.95s	20.52s	25.84s	71.70s	138.80s
mistralai/mistral-7b-instruct-v0.1	7.11s	23.05s	33.05s	79.21s	89.36s
openai/o3	8.08s	19.36s	27.07s	57.15s	76.89s
nvidia/nemotron-nano-9b-v2	12.05s	31.04s	36.70s	50.76s	65.31s
qwen/qwen3.6-flash	13.01s	33.57s	39.29s	47.21s	371.73s
google/gemini-2.5-pro	14.17s	24.82s	27.87s	80.22s	250.33s
deepseek/deepseek-r1-0528	16.52s	81.21s	94.87s	136.08s	285.69s
deepseek/deepseek-r1	19.86s	90.13s	153.13s	264.17s	364.40s
qwen/qwen3-14b	20.91s	43.46s	63.42s	225.55s	439.49s
nvidia/llama-3.3-nemotron-super-49b-v1.5	24.21s	82.63s	86.23s	99.39s	106.19s
deepseek/deepseek-v4-pro	26.42s	85.12s	101.25s	136.77s	190.30s
moonshotai/kimi-k2.6	35.30s	116.25s	160.11s	229.07s	411.39s
qwen/qwen3.6-plus	39.88s	69.37s	74.53s	90.19s	100.27s
qwen/qwen3.5-plus-02-15	48.19s	124.89s	143.16s	177.42s	1486.87s
qwen/qwen3-8b	59.45s	131.55s	144.22s	187.38s	233.05s
qwen/qwen3.5-27b	68.74s	273.44s	1145.40s	1651.28s	2172.22s

Output token efficiency

How many output tokens the model spent per detected ad. Lower is more concise (the model finds an ad and returns the JSON). Higher means the model is producing a lot of text the parser will discard, which costs you whether or not the answer is right.

Model	Total output tokens	Ads detected	Tokens / ad	Cost / TP
mistralai/mistral-medium-3.1	40,161	646	62	$0.0036
mistralai/codestral-2508	42,107	662	64	$0.0028
google/gemini-2.5-flash	67,460	925	73	$0.0027
google/gemini-3.1-flash-lite	60,147	815	74	$0.0018
openai/gpt-3.5-turbo	50,066	677	74	$0.0049
meta-llama/llama-3.3-70b-instruct	25,864	327	79	$0.0012
cohere/command-r-plus-08-2024	10,247	128	80	$0.0758
claude-sonnet-4-6	52,052	596	87	$0.0220
meta-llama/llama-3.1-8b-instruct	90,169	1015	89	$0.0004
meta-llama/llama-4-scout	48,083	540	89	$0.0010
mistralai/mistral-large-2512	108,068	1201	90	$0.0044
meta-llama/llama-4-maverick	38,620	415	93	$0.0011
claude-haiku-4-5-20251001	74,744	801	93	$0.0083
deepseek/deepseek-v3.2	33,405	348	96	$0.0031
google/gemini-2.5-flash-lite	89,161	917	97	$0.0008
google/gemma-4-31b-it	55,826	570	98	$0.0008
qwen/qwen3-235b-a22b-2507	59,253	560	106	$0.0005
cohere/command-a	55,492	522	106	$0.0199
openai/gpt-5.4	41,810	393	106	$0.0133
claude-opus-4-7	36,811	341	108	$0.0485
openai/gpt-5.4-mini	46,289	420	110	$0.0042
microsoft/phi-4	211,787	563	376	$0.0034
deepseek/deepseek-r1-distill-llama-70b	264,150	586	451	$0.0088
deepseek/deepseek-v4-flash	492,887	597	826	$0.0008
x-ai/grok-4.3	547,955	595	921	$0.0084
openai/gpt-5.5	310,390	330	941	$0.0361
openai/gpt-oss-120b	736,151	585	1258	$0.0004
deepseek/deepseek-r1-0528	1,197,928	871	1375	$0.0065
deepseek/deepseek-r1	817,850	580	1410	$0.0067
nvidia/nemotron-nano-9b-v2	1,220,752	556	2196	$0.0013
qwen/qwen3-14b	504,429	194	2600	$0.0021
google/gemini-3.5-flash	1,095,187	379	2890	$0.0189
google/gemini-2.5-pro	1,295,072	440	2943	$0.0188
qwen/qwen3.6-flash	1,566,281	416	3765	$0.0027
openai/o3	700,781	182	3850	$0.0223
nvidia/llama-3.3-nemotron-super-49b-v1.5	1,461,554	370	3950	$0.0020
deepseek/deepseek-v4-pro	1,148,041	213	5390	$0.0065
qwen/qwen3.6-plus	1,990,086	369	5393	$0.0054
qwen/qwen3.5-plus-02-15	3,059,957	373	8204	$0.0067
moonshotai/kimi-k2.6	2,323,592	226	10281	$0.0257
openai/o4-mini	954,370	33	28920	$0.1583
qwen/qwen3.5-27b	9,663,262	307	31476	$0.0199
qwen/qwen3-8b	2,583,129	6	430522	$0.2578

Trial variance (determinism check)

All trials run at temperature 0.0. If a model produces stable output you’d expect the F1 stdev across trials to be near zero. Higher numbers mean the model is non-deterministic even at temp=0. That’s fine to know, but means you cannot trust a single trial’s number for that model.

Model	Mean F1 stdev across episodes	Highest single-episode stdev
qwen/qwen3.6-plus	0.0399	0.1194
qwen/qwen3.5-plus-02-15	0.0340	0.0938
qwen/qwen3.6-flash	0.0869	0.2449
openai/gpt-5.5	0.0608	0.1432
google/gemini-3.5-flash	0.0216	0.0782
qwen/qwen3.5-27b	0.1606	0.3651
openai/gpt-5.4	0.0666	0.1193
google/gemini-2.5-pro	0.0405	0.1157
openai/gpt-5.4-mini	0.1123	0.2887
claude-opus-4-7	0.0431	0.1351
openai/o3	0.1498	0.4714
x-ai/grok-4.3	0.1048	0.2528
deepseek/deepseek-v4-flash	0.1229	0.3347
google/gemma-4-31b-it	0.0795	0.2739
deepseek/deepseek-r1	0.1426	0.3015
claude-sonnet-4-6	0.0416	0.1826
qwen/qwen3-235b-a22b-2507	0.0821	0.2653
openai/gpt-oss-120b	0.1380	0.3651
deepseek/deepseek-r1-0528	0.1424	0.2807
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.1348	0.3249
cohere/command-a	0.0402	0.1217
google/gemini-3.1-flash-lite	0.0389	0.2528
moonshotai/kimi-k2.6	0.1644	0.2739
meta-llama/llama-4-maverick	0.0162	0.1012
mistralai/mistral-medium-3.1	0.0467	0.1373
claude-haiku-4-5-20251001	0.0011	0.0116
deepseek/deepseek-v4-pro	0.2209	0.4333
mistralai/codestral-2508	0.0493	0.1493
google/gemini-2.5-flash-lite	0.0231	0.0523
deepseek/deepseek-v3.2	0.1908	0.5477
google/gemini-2.5-flash	0.0000	0.0000
openai/gpt-3.5-turbo	0.0129	0.0497
meta-llama/llama-3.3-70b-instruct	0.0814	0.2981
mistralai/mistral-large-2512	0.0162	0.0426
qwen/qwen3-14b	0.1663	0.3322
deepseek/deepseek-r1-distill-llama-70b	0.0974	0.3113
meta-llama/llama-4-scout	0.1091	0.3347
nvidia/nemotron-nano-9b-v2	0.1136	0.3651
meta-llama/llama-3.1-8b-instruct	0.1226	0.5477
cohere/command-r-plus-08-2024	0.0352	0.1547
openai/o4-mini	0.1255	0.2981
microsoft/phi-4	0.0614	0.3070
qwen/qwen3-8b	0.0106	0.1278
mistralai/mistral-7b-instruct-v0.1	0.0000	0.0000

Cross-model agreement

For each of the 171 (episode, window, trial-equivalent) entries, how many of the 44 active models predicted at least one ad? High-agreement windows are unambiguous ads (or unambiguously not ads). Low-agreement windows are where individual models disagree, and are candidates for ensemble voting if you want a cheap accuracy boost.

Models predicting an ad	Window count	Share
2 of 44	1	0.6%
4 of 44	8	4.7%
5 of 44	15	8.8%
6 of 44	10	5.8%
7 of 44	7	4.1%
8 of 44	6	3.5%
9 of 44	7	4.1%
10 of 44	9	5.3%
11 of 44	8	4.7%
12 of 44	7	4.1%
13 of 44	5	2.9%
14 of 44	2	1.2%
15 of 44	3	1.8%
16 of 44	4	2.3%
17 of 44	1	0.6%
18 of 44	1	0.6%
19 of 44	2	1.2%
26 of 44	2	1.2%
27 of 44	1	0.6%
29 of 44	3	1.8%
30 of 44	1	0.6%
32 of 44	1	0.6%
33 of 44	2	1.2%
35 of 44	2	1.2%
36 of 44	3	1.8%
37 of 44	6	3.5%
38 of 44	14	8.2%
39 of 44	16	9.4%
40 of 44	12	7.0%
41 of 44	9	5.3%
42 of 44	3	1.8%

Read this as: rows near the top are windows where the field disagrees (most models said no, a few said yes, usually false positives); rows near the bottom are windows where the field broadly agrees (typical of clear sponsor reads).

Per-model alignment with consensus

Same data, viewed per model. For each window, the majority is whether more than half of the 44 active models flagged an ad. Then for each model: did it vote with the majority or against it? Four buckets:

• with-yes: this model voted yes, majority also voted yes (likely true positive)
• with-no: this model voted no, majority also voted no (likely true negative)
• broke-yes: this model voted yes, majority voted no (likely false positive / hallucination)
• broke-no: this model voted no, majority voted yes (likely missed real ad)

Alignment rate is (with-yes + with-no) / total. High alignment means the model tracks the consensus; low alignment means it disagrees often, which could be brilliance or noise depending on whether its disagreements are also where its F1 wins or loses.

Model	with-yes	with-no	broke-yes	broke-no	Alignment
qwen/qwen3.5-plus-02-15	73	96	0	2	98.8%
x-ai/grok-4.3	73	96	0	2	98.8%
google/gemini-2.5-flash	74	94	2	1	98.2%
qwen/qwen3.5-27b	72	96	0	3	98.2%
openai/gpt-oss-120b	75	92	4	0	97.7%
openai/gpt-5.5	73	93	3	2	97.1%
nvidia/llama-3.3-nemotron-super-49b-v1.5	69	96	0	6	96.5%
claude-sonnet-4-6	67	96	0	8	95.3%
google/gemini-2.5-pro	75	88	8	0	95.3%
google/gemini-3.5-flash	67	96	0	8	95.3%
google/gemma-4-31b-it	73	90	6	2	95.3%
mistralai/mistral-medium-3.1	68	95	1	7	95.3%
qwen/qwen3.6-plus	68	95	1	7	95.3%
claude-haiku-4-5-20251001	69	93	3	6	94.7%
qwen/qwen3.6-flash	70	92	4	5	94.7%
openai/o3	63	96	0	12	93.0%
claude-opus-4-7	62	96	0	13	92.4%
meta-llama/llama-4-scout	73	85	11	2	92.4%
google/gemini-3.1-flash-lite	75	80	16	0	90.6%
meta-llama/llama-3.3-70b-instruct	64	89	7	11	89.5%
deepseek/deepseek-v4-flash	75	77	19	0	88.9%
nvidia/nemotron-nano-9b-v2	73	78	18	2	88.3%
meta-llama/llama-4-maverick	72	78	18	3	87.7%
deepseek/deepseek-r1	75	74	22	0	87.1%
meta-llama/llama-3.1-8b-instruct	70	78	18	5	86.5%
deepseek/deepseek-v4-pro	61	82	14	14	83.6%
google/gemini-2.5-flash-lite	75	68	28	0	83.6%
mistralai/codestral-2508	69	71	25	6	81.9%
openai/gpt-5.4	74	66	30	1	81.9%
qwen/qwen3-14b	65	73	23	10	80.7%
openai/gpt-5.4-mini	73	59	37	2	77.2%
cohere/command-a	71	56	40	4	74.3%
cohere/command-r-plus-08-2024	25	96	0	50	70.8%
openai/o4-mini	25	96	0	50	70.8%
deepseek/deepseek-v3.2	53	63	33	22	67.8%
mistralai/mistral-large-2512	74	41	55	1	67.3%
deepseek/deepseek-r1-0528	71	34	62	4	61.4%
qwen/qwen3-235b-a22b-2507	75	30	66	0	61.4%
openai/gpt-3.5-turbo	75	28	68	0	60.2%
qwen/qwen3-8b	2	96	0	73	57.3%
mistralai/mistral-7b-instruct-v0.1	0	96	0	75	56.1%
moonshotai/kimi-k2.6	48	42	54	27	52.6%
deepseek/deepseek-r1-distill-llama-70b	66	8	88	9	43.3%
microsoft/phi-4	55	14	82	20	40.4%

Detection rate by ad characteristic

Aggregate detection rates often hide systematic blind spots. Below: for each model, what fraction of truth ads in each bucket were detected (matched at IoU >= 0.5).

By ad length

Truth ads bucketed by duration: short (<30s), medium (30-90s), long (>=90s). Cell values are detection rate (fraction of truth ads in that bucket the model caught), with the sample size n so a misleading 1.00 on a 2-ad bucket doesn’t get over-weighted. Models that systematically miss short ads usually fail on network-inserted brand-tagline spots; missing long ads is rarer and usually means the model gave up before processing the full window.

Model	long (>=90s)	medium (30-90s)	short (<30s)
claude-haiku-4-5-20251001	0.36 (n=140)	0.80 (n=75)	0.88 (n=40)
claude-opus-4-7	0.61 (n=140)	0.75 (n=75)	0.50 (n=40)
claude-sonnet-4-6	0.54 (n=140)	0.77 (n=75)	0.68 (n=40)
cohere/command-a	0.41 (n=140)	0.63 (n=75)	0.75 (n=40)
cohere/command-r-plus-08-2024	0.19 (n=140)	0.04 (n=75)	0.12 (n=40)
deepseek/deepseek-r1	0.56 (n=140)	0.76 (n=75)	0.80 (n=40)
deepseek/deepseek-r1-0528	0.51 (n=140)	0.73 (n=75)	0.78 (n=40)
deepseek/deepseek-r1-distill-llama-70b	0.26 (n=140)	0.29 (n=75)	0.62 (n=40)
deepseek/deepseek-v3.2	0.35 (n=140)	0.31 (n=75)	0.28 (n=40)
deepseek/deepseek-v4-flash	0.61 (n=140)	0.81 (n=75)	0.78 (n=40)
deepseek/deepseek-v4-pro	0.49 (n=140)	0.28 (n=75)	0.20 (n=40)
google/gemini-2.5-flash	0.29 (n=140)	0.67 (n=75)	0.88 (n=40)
google/gemini-2.5-flash-lite	0.35 (n=140)	0.76 (n=75)	0.88 (n=40)
google/gemini-2.5-pro	0.84 (n=140)	0.79 (n=75)	0.72 (n=40)
google/gemini-3.1-flash-lite	0.34 (n=140)	0.92 (n=75)	0.88 (n=40)
google/gemini-3.5-flash	0.72 (n=140)	0.76 (n=75)	0.70 (n=40)
google/gemma-4-31b-it	0.56 (n=140)	0.73 (n=75)	0.60 (n=40)
meta-llama/llama-3.1-8b-instruct	0.12 (n=140)	0.28 (n=75)	0.42 (n=40)
meta-llama/llama-3.3-70b-instruct	0.30 (n=140)	0.32 (n=75)	0.53 (n=40)
meta-llama/llama-4-maverick	0.50 (n=140)	0.69 (n=75)	0.38 (n=40)
meta-llama/llama-4-scout	0.22 (n=140)	0.36 (n=75)	0.57 (n=40)
microsoft/phi-4	0.04 (n=140)	0.15 (n=75)	0.12 (n=40)
mistralai/codestral-2508	0.39 (n=140)	0.57 (n=75)	0.45 (n=40)
mistralai/mistral-7b-instruct-v0.1	0.00 (n=140)	0.00 (n=75)	0.00 (n=40)
mistralai/mistral-large-2512	0.31 (n=140)	0.73 (n=75)	0.75 (n=40)
mistralai/mistral-medium-3.1	0.30 (n=140)	0.69 (n=75)	0.72 (n=40)
moonshotai/kimi-k2.6	0.30 (n=140)	0.52 (n=75)	0.25 (n=40)
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.31 (n=140)	0.57 (n=75)	0.60 (n=40)
nvidia/nemotron-nano-9b-v2	0.14 (n=140)	0.43 (n=75)	0.35 (n=40)
openai/gpt-3.5-turbo	0.31 (n=140)	0.40 (n=75)	0.75 (n=40)
openai/gpt-5.4	0.80 (n=140)	0.73 (n=75)	0.57 (n=40)
openai/gpt-5.4-mini	0.74 (n=140)	0.63 (n=75)	0.75 (n=40)
openai/gpt-5.5	0.74 (n=140)	0.77 (n=75)	0.57 (n=40)
openai/gpt-oss-120b	0.46 (n=140)	0.71 (n=75)	0.75 (n=40)
openai/o3	0.66 (n=140)	0.44 (n=75)	0.30 (n=40)
openai/o4-mini	0.03 (n=140)	0.05 (n=75)	0.10 (n=40)
qwen/qwen3-14b	0.23 (n=140)	0.21 (n=75)	0.30 (n=40)
qwen/qwen3-235b-a22b-2507	0.44 (n=140)	0.65 (n=75)	0.88 (n=40)
qwen/qwen3-8b	0.01 (n=140)	0.00 (n=75)	0.00 (n=40)
qwen/qwen3.5-27b	0.59 (n=140)	0.69 (n=75)	0.68 (n=40)
qwen/qwen3.5-plus-02-15	0.86 (n=116)	0.81 (n=72)	0.62 (n=40)
qwen/qwen3.6-flash	0.87 (n=140)	0.73 (n=75)	0.70 (n=40)
qwen/qwen3.6-plus	0.84 (n=140)	0.83 (n=75)	0.62 (n=40)
x-ai/grok-4.3	0.58 (n=140)	0.91 (n=75)	0.72 (n=40)

By ad position

Truth ads bucketed by where they fall in the episode: pre-roll (first 10%), mid-roll (10-90%), post-roll (last 10%). Cell values are the same detection-rate-with-n format as ad length. A common failure pattern in our data: most models detect pre-roll and mid-roll reliably and miss post-roll, because the prompt windows near the end often catch the model mid-reasoning or with fewer transition phrases to anchor on.

Model	pre-roll (<10%)	mid-roll (10-90%)	post-roll (>90%)
claude-haiku-4-5-20251001	0.53 (n=75)	0.56 (n=125)	0.64 (n=55)
claude-opus-4-7	0.65 (n=75)	0.56 (n=125)	0.76 (n=55)
claude-sonnet-4-6	0.77 (n=75)	0.58 (n=125)	0.55 (n=55)
cohere/command-a	0.52 (n=75)	0.52 (n=125)	0.55 (n=55)
cohere/command-r-plus-08-2024	0.05 (n=75)	0.23 (n=125)	0.02 (n=55)
deepseek/deepseek-r1	0.61 (n=75)	0.69 (n=125)	0.64 (n=55)
deepseek/deepseek-r1-0528	0.59 (n=75)	0.63 (n=125)	0.62 (n=55)
deepseek/deepseek-r1-distill-llama-70b	0.31 (n=75)	0.33 (n=125)	0.35 (n=55)
deepseek/deepseek-v3.2	0.41 (n=75)	0.37 (n=125)	0.11 (n=55)
deepseek/deepseek-v4-flash	0.63 (n=75)	0.75 (n=125)	0.65 (n=55)
deepseek/deepseek-v4-pro	0.27 (n=75)	0.47 (n=125)	0.35 (n=55)
google/gemini-2.5-flash	0.47 (n=75)	0.48 (n=125)	0.55 (n=55)
google/gemini-2.5-flash-lite	0.59 (n=75)	0.54 (n=125)	0.55 (n=55)
google/gemini-2.5-pro	0.79 (n=75)	0.85 (n=125)	0.73 (n=55)
google/gemini-3.1-flash-lite	0.60 (n=75)	0.58 (n=125)	0.62 (n=55)
google/gemini-3.5-flash	0.67 (n=75)	0.77 (n=125)	0.73 (n=55)
google/gemma-4-31b-it	0.59 (n=75)	0.62 (n=125)	0.67 (n=55)
meta-llama/llama-3.1-8b-instruct	0.23 (n=75)	0.23 (n=125)	0.16 (n=55)
meta-llama/llama-3.3-70b-instruct	0.23 (n=75)	0.40 (n=125)	0.36 (n=55)
meta-llama/llama-4-maverick	0.49 (n=75)	0.60 (n=125)	0.45 (n=55)
meta-llama/llama-4-scout	0.27 (n=75)	0.40 (n=125)	0.20 (n=55)
microsoft/phi-4	0.17 (n=75)	0.02 (n=125)	0.11 (n=55)
mistralai/codestral-2508	0.36 (n=75)	0.48 (n=125)	0.53 (n=55)
mistralai/mistral-7b-instruct-v0.1	0.00 (n=75)	0.00 (n=125)	0.00 (n=55)
mistralai/mistral-large-2512	0.52 (n=75)	0.51 (n=125)	0.45 (n=55)
mistralai/mistral-medium-3.1	0.51 (n=75)	0.44 (n=125)	0.55 (n=55)
moonshotai/kimi-k2.6	0.29 (n=75)	0.31 (n=125)	0.55 (n=55)
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.37 (n=75)	0.46 (n=125)	0.44 (n=55)
nvidia/nemotron-nano-9b-v2	0.15 (n=75)	0.26 (n=125)	0.38 (n=55)
openai/gpt-3.5-turbo	0.31 (n=75)	0.48 (n=125)	0.36 (n=55)
openai/gpt-5.4	0.75 (n=75)	0.78 (n=125)	0.65 (n=55)
openai/gpt-5.4-mini	0.60 (n=75)	0.83 (n=125)	0.56 (n=55)
openai/gpt-5.5	0.73 (n=75)	0.73 (n=125)	0.71 (n=55)
openai/gpt-oss-120b	0.53 (n=75)	0.61 (n=125)	0.56 (n=55)
openai/o3	0.44 (n=75)	0.59 (n=125)	0.55 (n=55)
openai/o4-mini	0.01 (n=75)	0.06 (n=125)	0.05 (n=55)
qwen/qwen3-14b	0.13 (n=75)	0.29 (n=125)	0.25 (n=55)
qwen/qwen3-235b-a22b-2507	0.56 (n=75)	0.59 (n=125)	0.55 (n=55)
qwen/qwen3-8b	0.00 (n=75)	0.01 (n=125)	0.00 (n=55)
qwen/qwen3.5-27b	0.56 (n=75)	0.67 (n=125)	0.65 (n=55)
qwen/qwen3.5-plus-02-15	0.82 (n=66)	0.84 (n=110)	0.71 (n=52)
qwen/qwen3.6-flash	0.71 (n=75)	0.88 (n=125)	0.76 (n=55)
qwen/qwen3.6-plus	0.75 (n=75)	0.88 (n=125)	0.71 (n=55)
x-ai/grok-4.3	0.67 (n=75)	0.74 (n=125)	0.65 (n=55)

Quick Comparison

One row per model, one column per episode. The headline columns (F1, Cost/ep, p50) summarize across all episodes; the per-episode columns let you see whether a model’s average hides wide swings (a model that scores well overall might still bomb on a specific genre). The right-most F1 stdev column averages the per-trial standard deviations across episodes; high values mean the model isn’t deterministic at temperature 0.0, so its single-trial F1 number is noisy.

Model	F1	Cost/ep	p50	ep-crime-junkie-8ce498f299d7	ep-daily-gist-chicago-70a82fe93a5c	ep-daily-tech-news-show-b576979e1fe8	ep-daily-tech-news-show-c1904b8605f7	ep-drink-champs-30c9a2d49f13	ep-glt1412515089-373d5ba5007b	ep-it-s-a-thing-e339179dfad6	ep-on-air-with-dan-and-alex2-574e4f303730	ep-security-now-audio-2850b24903b2	ep-the-brilliant-idiots-0bb9bf634c8e	ep-the-tim-dillon-show-f62bd5fa1cfe	ep-tosh-show-5f6894439bb6	ep-ai-cloud-essentials-e8dc897fbd6b (no-ad)	ep-oxide-and-friends-ce789ff5b62e (no-ad)	F1 stdev
qwen/qwen3.6-plus	0.693	$1.1119	39.9s	0.933	0.667	0.914	0.457	0.706	0.600	0.667	0.773	0.481	0.798	0.615	0.703	PASS	PASS	0.040
qwen/qwen3.5-plus-02-15	0.679	$1.2346	48.2s	1.000	0.600	0.857	0.518	0.711	0.625	0.667	0.590	0.476	0.771	0.636	0.696	PASS	PASS	0.034
qwen/qwen3.6-flash	0.660	$0.5461	13.0s	1.000	0.700	0.852	0.492	0.628	0.518	0.667	0.682	0.514	0.800	0.507	0.564	FAIL (1 FP)	PASS	0.087
openai/gpt-5.5	0.647	$6.6806	6.4s	0.898	0.500	0.886	0.547	0.642	0.636	0.667	0.571	0.505	0.776	0.546	0.587	FAIL (1 FP)	PASS	0.061
google/gemini-3.5-flash	0.625	$3.5200	5.2s	0.914	0.400	0.857	0.500	0.496	0.676	0.667	0.571	0.476	0.857	0.592	0.490	PASS	PASS	0.022
qwen/qwen3.5-27b	0.622	$3.2163	68.7s	0.892	0.727	0.743	0.506	0.494	0.601	0.600	0.634	0.414	0.671	0.537	0.647	PASS	PASS	0.161
openai/gpt-5.4	0.620	$2.5345	1.8s	0.933	0.500	0.892	0.518	0.571	0.506	0.613	0.747	0.495	0.516	0.586	0.569	FAIL (1 FP)	FAIL (1 FP)	0.067
google/gemini-2.5-pro	0.613	$3.8631	14.2s	0.971	0.400	0.864	0.448	0.679	0.646	0.667	0.543	0.451	0.510	0.569	0.603	FAIL (1 FP)	FAIL (1 FP)	0.041
openai/gpt-5.4-mini	0.584	$0.7644	1.2s	0.883	0.833	0.943	0.468	0.509	0.494	0.500	0.610	0.491	0.353	0.373	0.547	FAIL (1 FP)	FAIL (1 FP)	0.112
claude-opus-4-7	0.576	$7.8054	2.2s	0.857	0.400	0.863	0.231	0.334	0.600	0.667	0.595	0.520	0.733	0.592	0.524	PASS	PASS	0.043
openai/o3	0.567	$3.0485	8.1s	0.876	0.000	0.848	0.472	0.742	0.664	0.333	0.687	0.644	0.698	0.372	0.464	PASS	PASS	0.150
x-ai/grok-4.3	0.506	$1.4987	3.9s	0.938	0.533	0.507	0.179	0.208	0.572	0.433	0.472	0.486	0.771	0.393	0.585	PASS	PASS	0.105
deepseek/deepseek-v4-flash	0.475	$0.1346	3.7s	0.718	0.587	0.445	0.310	0.218	0.642	0.337	0.535	0.465	0.569	0.298	0.574	FAIL (1 FP)	PASS	0.123
google/gemma-4-31b-it	0.474	$0.1291	2.2s	0.879	0.480	0.811	0.119	0.158	0.645	0.467	0.514	0.496	0.585	0.350	0.182	FAIL (1 FP)	PASS	0.080
deepseek/deepseek-r1	0.469	$1.1157	19.9s	0.732	0.693	0.658	0.158	0.268	0.630	0.313	0.555	0.462	0.401	0.327	0.435	FAIL (1 FP)	FAIL (1 FP)	0.143
claude-sonnet-4-6	0.446	$3.5376	1.4s	0.889	0.800	0.407	0.237	0.275	0.400	0.000	0.444	0.516	0.836	0.179	0.375	PASS	PASS	0.042
qwen/qwen3-235b-a22b-2507	0.433	$0.0735	2.3s	0.774	0.853	0.756	0.142	0.033	0.540	0.000	0.517	0.460	0.342	0.421	0.362	FAIL (2 FP)	FAIL (6 FP)	0.082
openai/gpt-oss-120b	0.432	$0.0643	3.0s	0.819	0.600	0.270	0.201	0.058	0.625	0.227	0.457	0.469	0.630	0.387	0.443	FAIL (1 FP)	PASS	0.138
deepseek/deepseek-r1-0528	0.420	$1.0181	16.5s	0.717	0.647	0.700	0.184	0.125	0.379	0.404	0.514	0.281	0.161	0.283	0.643	FAIL (27 FP)	FAIL (12 FP)	0.142
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.411	$0.2159	24.2s	0.888	0.667	0.299	0.139	0.045	0.596	0.233	0.431	0.587	0.478	0.171	0.399	PASS	PASS	0.135
cohere/command-a	0.408	$2.6688	3.8s	0.911	0.400	0.500	0.298	0.031	0.507	0.400	0.503	0.368	0.247	0.200	0.533	FAIL (3 FP)	PASS	0.040
google/gemini-3.1-flash-lite	0.402	$0.2762	0.8s	0.771	0.800	0.473	0.076	0.039	0.527	0.433	0.400	0.413	0.338	0.183	0.375	FAIL (1 FP)	PASS	0.039
moonshotai/kimi-k2.6	0.400	$2.3369	35.3s	0.547	0.100	0.914	0.600	0.053	0.469	0.200	0.734	0.196	0.538	0.184	0.267	FAIL (1 FP)	FAIL (4 FP)	0.164
meta-llama/llama-4-maverick	0.398	$0.1508	1.1s	1.000	0.000	0.771	0.204	0.273	0.507	0.000	0.571	0.496	0.390	0.167	0.400	FAIL (1 FP)	PASS	0.016
mistralai/mistral-medium-3.1	0.368	$0.4380	0.9s	0.397	0.667	0.162	0.095	0.046	0.671	0.000	0.500	0.640	0.591	0.223	0.421	PASS	PASS	0.047
claude-haiku-4-5-20251001	0.367	$1.2017	1.2s	0.500	0.800	0.235	0.073	0.042	0.571	0.000	0.500	0.551	0.600	0.154	0.375	PASS	PASS	0.001
deepseek/deepseek-v4-pro	0.367	$0.6376	26.4s	0.728	0.133	0.418	0.350	0.262	0.449	0.267	0.257	0.498	0.167	0.565	0.305	PASS	PASS	0.221
mistralai/codestral-2508	0.314	$0.3241	0.7s	0.520	0.667	0.379	0.172	0.033	0.231	0.000	0.469	0.374	0.187	0.178	0.564	PASS	PASS	0.049
google/gemini-2.5-flash-lite	0.302	$0.1104	0.9s	0.462	0.667	0.326	0.069	0.084	0.413	0.000	0.400	0.280	0.432	0.133	0.364	FAIL (1 FP)	PASS	0.023
deepseek/deepseek-v3.2	0.298	$0.2564	2.7s	0.590	0.300	0.327	0.140	0.176	0.518	0.400	0.300	0.528	0.100	0.057	0.140	PASS	FAIL (2 FP)	0.191
google/gemini-2.5-flash	0.293	$0.3435	1.0s	0.462	0.800	0.267	0.071	0.037	0.500	0.000	0.444	0.455	0.125	0.154	0.200	PASS	PASS	0.000
openai/gpt-3.5-turbo	0.286	$0.5091	1.3s	0.978	0.420	0.222	0.364	0.038	0.254	0.000	0.400	0.217	0.220	0.125	0.189	FAIL (3 FP)	FAIL (10 FP)	0.013
meta-llama/llama-3.3-70b-instruct	0.274	$0.1008	1.5s	0.502	0.100	0.327	0.000	0.000	0.567	0.000	0.133	0.512	0.489	0.308	0.347	PASS	PASS	0.081
mistralai/mistral-large-2512	0.249	$0.5599	2.6s	0.492	0.648	0.250	0.074	0.033	0.253	0.000	0.444	0.209	0.193	0.044	0.353	PASS	PASS	0.016
qwen/qwen3-14b	0.248	$0.1261	20.9s	0.190	0.000	0.518	0.100	0.045	0.434	0.000	0.330	0.336	0.372	0.233	0.412	PASS	FAIL (1 FP)	0.166
deepseek/deepseek-r1-distill-llama-70b	0.241	$0.7331	2.0s	0.590	0.440	0.057	0.205	0.000	0.265	0.000	0.343	0.262	0.222	0.163	0.340	FAIL (2 FP)	FAIL (10 FP)	0.097
meta-llama/llama-4-scout	0.235	$0.0807	0.8s	0.253	0.587	0.242	0.130	0.000	0.336	0.000	0.094	0.520	0.349	0.053	0.258	PASS	PASS	0.109
nvidia/nemotron-nano-9b-v2	0.221	$0.0813	12.0s	0.334	0.600	0.274	0.069	0.044	0.242	0.000	0.361	0.238	0.205	0.044	0.240	FAIL (1 FP)	PASS	0.114
meta-llama/llama-3.1-8b-instruct	0.198	$0.0207	0.8s	0.164	0.700	0.231	0.029	0.017	0.183	0.400	0.284	0.133	0.000	0.049	0.185	PASS	PASS	0.123
cohere/command-r-plus-08-2024	0.102	$2.5787	1.0s	0.000	0.000	0.000	0.313	0.000	0.000	0.000	0.000	0.569	0.000	0.057	0.289	PASS	PASS	0.035
openai/o4-mini	0.075	$1.8999	6.9s	0.000	0.000	0.147	0.067	0.040	0.080	0.000	0.133	0.114	0.180	0.000	0.133	PASS	PASS	0.125
microsoft/phi-4	0.055	$0.0713	2.3s	0.157	0.000	0.000	0.056	0.058	0.000	0.000	0.213	0.033	0.067	0.079	0.000	FAIL (3 FP)	FAIL (15 FP)	0.061
qwen/qwen3-8b	0.005	$0.2578	59.4s	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.057	0.000	PASS	PASS	0.011
mistralai/mistral-7b-instruct-v0.1	0.000	$0.0285	7.1s	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	0.000	PASS	PASS	0.000

---

Detailed Results

Per-Model Detail

Full per-model profile: F1 averaged across episodes, total cost per episode at current pricing, p50 / p95 latency, JSON compliance, parse-failure rate, the distribution of extraction methods the parser had to use, and verbosity / truncation telemetry. The Extraction methods list shows how often each route was hit. json_array_direct is the cleanest; the rest are recovery paths. The verbosity row flags models that emit long reason fields or run out of token budget mid-response. Ordered by F1 descending so the best performers appear first.

qwen/qwen3.6-plus

• F1 (avg across episodes): 0.693
• Total cost / episode: $1.1119
• p50 / p95 latency: 39.88s / 74.53s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 855
• Verbosity: 813/855 calls over 1024 output tokens (95.1%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

qwen/qwen3.5-plus-02-15

• F1 (avg across episodes): 0.679
• Total cost / episode: $1.2346
• p50 / p95 latency: 48.19s / 143.16s
• JSON compliance: 1.00
• JSON mode: native (100% native, 852 calls)
• Parse failure rate: 0.1%
• Extraction methods: json_array_direct: 851, parse_failure: 1
• Verbosity: 749/852 calls over 1024 output tokens (87.9%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 384
• Extra keys observed: end_text, sponsor

qwen/qwen3.6-flash

• F1 (avg across episodes): 0.660
• Total cost / episode: $0.5461
• p50 / p95 latency: 13.01s / 39.29s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 855
• Verbosity: 585/855 calls over 1024 output tokens (68.4%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

openai/gpt-5.5

• F1 (avg across episodes): 0.647
• Total cost / episode: $6.6806
• p50 / p95 latency: 6.38s / 24.05s
• JSON compliance: 0.87
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.1%
• Extraction methods: json_object_no_ads: 494, json_object_single_ad: 360, parse_failure: 1
• Verbosity: 76/855 calls over 1024 output tokens (8.9%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 332
• Extra keys observed: end_text, sponsor

google/gemini-3.5-flash

• F1 (avg across episodes): 0.625
• Total cost / episode: $3.5200
• p50 / p95 latency: 5.23s / 11.13s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 850, json_object_single_ad_truncated: 1, regex_json_array: 4
• Verbosity: 519/855 calls over 1024 output tokens (60.7%); 1 hit max_tokens (0.1%); 1 salvaged from truncated JSON (0.1%)

qwen/qwen3.5-27b

• F1 (avg across episodes): 0.622
• Total cost / episode: $3.2163
• p50 / p95 latency: 68.74s / 1145.40s
• JSON compliance: 0.85
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 14.2%
• Extraction methods: json_array_direct: 633, json_object_no_ads: 89, json_object_single_ad: 12, parse_failure: 121
• Verbosity: 798/855 calls over 1024 output tokens (93.3%); 108 hit max_tokens (12.6%); 0 salvaged from truncated JSON (0.0%)

openai/gpt-5.4

• F1 (avg across episodes): 0.620
• Total cost / episode: $2.5345
• p50 / p95 latency: 1.82s / 3.12s
• JSON compliance: 0.81
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_object_no_ads: 326, json_object_single_ad: 529
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 398
• Extra keys observed: end_text, sponsor

google/gemini-2.5-pro

• F1 (avg across episodes): 0.613
• Total cost / episode: $3.8631
• p50 / p95 latency: 14.17s / 27.87s
• JSON compliance: 0.97
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 1.4%
• Extraction methods: json_array_direct: 818, parse_failure: 12, regex_json_array: 25
• Verbosity: 673/855 calls over 1024 output tokens (78.7%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 389
• Extra keys observed: end_text, sponsor

openai/gpt-5.4-mini

• F1 (avg across episodes): 0.584
• Total cost / episode: $0.7644
• p50 / p95 latency: 1.18s / 2.24s
• JSON compliance: 0.81
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_object_ads_key: 2, json_object_no_ads: 300, json_object_segments_key: 2, json_object_single_ad: 551
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

claude-opus-4-7

• F1 (avg across episodes): 0.576
• Total cost / episode: $7.8054
• p50 / p95 latency: 2.17s / 4.21s
• JSON compliance: 1.00
• JSON mode: prompt-inject (0% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 850, regex_json_array: 5
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 341
• Extra keys observed: end_text, sponsor

openai/o3

• F1 (avg across episodes): 0.567
• Total cost / episode: $3.0485
• p50 / p95 latency: 8.08s / 27.07s
• JSON compliance: 0.92
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.6%
• Extraction methods: json_object_ads_key: 34, json_object_no_ads: 621, json_object_segments_key: 12, json_object_single_ad: 183, parse_failure: 5
• Verbosity: 246/855 calls over 1024 output tokens (28.8%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 132
• Extra keys observed: end_text, sponsor

x-ai/grok-4.3

• F1 (avg across episodes): 0.506
• Total cost / episode: $1.4987
• p50 / p95 latency: 3.88s / 12.56s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.1%
• Extraction methods: json_array_direct: 854, parse_failure: 1
• Verbosity: 165/855 calls over 1024 output tokens (19.3%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

deepseek/deepseek-v4-flash

• F1 (avg across episodes): 0.475
• Total cost / episode: $0.1346
• p50 / p95 latency: 3.67s / 28.76s
• JSON compliance: 0.81
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 2.0%
• Extraction methods: json_array_direct: 102, json_object_ads_key: 435, json_object_no_ads: 21, json_object_segments_key: 7, json_object_single_ad: 272, parse_failure: 17, regex_json_array: 1
• Verbosity: 181/855 calls over 1024 output tokens (21.2%); 2 hit max_tokens (0.2%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 631
• Extra keys observed: end_text, sponsor

google/gemma-4-31b-it

• F1 (avg across episodes): 0.474
• Total cost / episode: $0.1291
• p50 / p95 latency: 2.21s / 19.39s
• JSON compliance: 0.85
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.1%
• Extraction methods: bracket_fallback: 1, json_object_ads_key: 455, json_object_no_ads: 216, json_object_single_ad: 180, json_object_single_ad_truncated: 2, parse_failure: 1
• Verbosity: 3/855 calls over 1024 output tokens (0.4%); 3 hit max_tokens (0.4%); 2 salvaged from truncated JSON (0.2%)
• Schema violations: 532
• Extra keys observed: end_text, sponsor

deepseek/deepseek-r1

• F1 (avg across episodes): 0.469
• Total cost / episode: $1.1157
• p50 / p95 latency: 19.86s / 153.13s
• JSON compliance: 0.97
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.7%
• Extraction methods: json_array_direct: 760, json_object_ads_key: 2, json_object_no_ads: 17, json_object_segments_key: 7, json_object_single_ad: 44, markdown_code_block: 17, parse_failure: 6, regex_json_array: 2
• Verbosity: 152/855 calls over 1024 output tokens (17.8%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 486
• Extra keys observed: end_text, sponsor

claude-sonnet-4-6

• F1 (avg across episodes): 0.446
• Total cost / episode: $3.5376
• p50 / p95 latency: 1.44s / 6.04s
• JSON compliance: 0.96
• JSON mode: prompt-inject (0% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 783, markdown_code_block: 57, regex_json_array: 15
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 461
• Extra keys observed: end_text, sponsor

qwen/qwen3-235b-a22b-2507

• F1 (avg across episodes): 0.433
• Total cost / episode: $0.0735
• p50 / p95 latency: 2.28s / 7.75s
• JSON compliance: 0.79
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 173, json_object_ads_key: 1, json_object_no_ads: 94, json_object_single_ad: 587
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

openai/gpt-oss-120b

• F1 (avg across episodes): 0.432
• Total cost / episode: $0.0643
• p50 / p95 latency: 2.96s / 34.40s
• JSON compliance: 0.70
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 17.4%
• Extraction methods: json_array_direct: 68, json_object_ads_key: 235, json_object_no_ads: 186, json_object_single_ad: 188, json_object_single_ad_truncated: 6, parse_failure: 149, regex_json_array: 23
• Verbosity: 162/855 calls over 1024 output tokens (18.9%); 2 hit max_tokens (0.2%); 6 salvaged from truncated JSON (0.7%)

deepseek/deepseek-r1-0528

• F1 (avg across episodes): 0.420
• Total cost / episode: $1.0181
• p50 / p95 latency: 16.52s / 94.87s
• JSON compliance: 0.88
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 9.7%
• Extraction methods: json_array_direct: 700, json_object_ads_key: 34, json_object_no_ads: 3, json_object_single_ad: 30, json_object_single_ad_truncated: 3, markdown_code_block: 2, parse_failure: 83
• Verbosity: 341/855 calls over 1024 output tokens (39.9%); 38 hit max_tokens (4.4%); 3 salvaged from truncated JSON (0.4%)
• Schema violations: 694
• Extra keys observed: end_text, sponsor

nvidia/llama-3.3-nemotron-super-49b-v1.5

• F1 (avg across episodes): 0.411
• Total cost / episode: $0.2159
• p50 / p95 latency: 24.21s / 86.23s
• JSON compliance: 0.71
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 14.5%
• Extraction methods: json_array_direct: 441, json_object_single_ad_truncated: 3, markdown_code_block: 257, parse_failure: 124, regex_json_array: 30
• Verbosity: 515/855 calls over 1024 output tokens (60.2%); 58 hit max_tokens (6.8%); 3 salvaged from truncated JSON (0.4%)
• Schema violations: 266
• Extra keys observed: end_text, sponsor

cohere/command-a

• F1 (avg across episodes): 0.408
• Total cost / episode: $2.6688
• p50 / p95 latency: 3.78s / 11.46s
• JSON compliance: 0.71
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_object_no_ads: 29, json_object_single_ad: 826
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 478
• Extra keys observed: end_text, sponsor

google/gemini-3.1-flash-lite

• F1 (avg across episodes): 0.402
• Total cost / episode: $0.2762
• p50 / p95 latency: 0.79s / 1.44s
• JSON compliance: 0.96
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 799, regex_json_array: 56
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

moonshotai/kimi-k2.6

• F1 (avg across episodes): 0.400
• Total cost / episode: $2.3369
• p50 / p95 latency: 35.30s / 160.11s
• JSON compliance: 0.57
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 28.0%
• Extraction methods: json_array_direct: 68, json_object_ads_key: 35, json_object_no_ads: 109, json_object_segments_key: 2, json_object_single_ad: 397, markdown_code_block: 5, parse_failure: 239
• Verbosity: 767/855 calls over 1024 output tokens (89.7%); 113 hit max_tokens (13.2%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 234
• Extra keys observed: end_text, sponsor

meta-llama/llama-4-maverick

• F1 (avg across episodes): 0.398
• Total cost / episode: $0.1508
• p50 / p95 latency: 1.06s / 2.38s
• JSON compliance: 0.81
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_object_no_ads: 315, json_object_single_ad: 540
• Verbosity: 3/855 calls over 1024 output tokens (0.4%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 372
• Extra keys observed: end_text, sponsor

mistralai/mistral-medium-3.1

• F1 (avg across episodes): 0.368
• Total cost / episode: $0.4380
• p50 / p95 latency: 0.91s / 6.18s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 854, json_object_single_ad_truncated: 1
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 1 salvaged from truncated JSON (0.1%)
• Schema violations: 608
• Extra keys observed: end_text, sponsor

claude-haiku-4-5-20251001

• F1 (avg across episodes): 0.367
• Total cost / episode: $1.2017
• p50 / p95 latency: 1.17s / 4.06s
• JSON compliance: 0.60
• JSON mode: prompt-inject (0% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: markdown_code_block: 855
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 711
• Extra keys observed: end_text, sponsor

deepseek/deepseek-v4-pro

• F1 (avg across episodes): 0.367
• Total cost / episode: $0.6376
• p50 / p95 latency: 26.42s / 101.25s
• JSON compliance: 0.87
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 2.7%
• Extraction methods: json_array_direct: 229, json_object_ads_key: 63, json_object_no_ads: 131, json_object_segments_key: 314, json_object_single_ad: 86, json_object_single_ad_truncated: 2, markdown_code_block: 5, parse_failure: 23, regex_json_array: 2
• Verbosity: 433/855 calls over 1024 output tokens (50.6%); 15 hit max_tokens (1.8%); 2 salvaged from truncated JSON (0.2%)

mistralai/codestral-2508

• F1 (avg across episodes): 0.314
• Total cost / episode: $0.3241
• p50 / p95 latency: 0.73s / 2.24s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 855
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 622
• Extra keys observed: end_text, sponsor

google/gemini-2.5-flash-lite

• F1 (avg across episodes): 0.302
• Total cost / episode: $0.1104
• p50 / p95 latency: 0.92s / 3.23s
• JSON compliance: 0.97
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 2.2%
• Extraction methods: json_array_direct: 793, json_object_single_ad_truncated: 43, parse_failure: 19
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 43 salvaged from truncated JSON (5.0%)

deepseek/deepseek-v3.2

• F1 (avg across episodes): 0.298
• Total cost / episode: $0.2564
• p50 / p95 latency: 2.67s / 7.32s
• JSON compliance: 0.88
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 501, json_object_ads_key: 16, json_object_no_ads: 3, json_object_single_ad: 335
• Verbosity: 1/855 calls over 1024 output tokens (0.1%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 263
• Extra keys observed: end_text, sponsor

google/gemini-2.5-flash

• F1 (avg across episodes): 0.293
• Total cost / episode: $0.3435
• p50 / p95 latency: 0.99s / 3.59s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 855
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 710
• Extra keys observed: end_text, sponsor

openai/gpt-3.5-turbo

• F1 (avg across episodes): 0.286
• Total cost / episode: $0.5091
• p50 / p95 latency: 1.26s / 1.95s
• JSON compliance: 0.71
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.6%
• Extraction methods: json_object_no_ads: 50, json_object_single_ad: 800, parse_failure: 5
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 593
• Extra keys observed: end_text, sponsor

meta-llama/llama-3.3-70b-instruct

• F1 (avg across episodes): 0.274
• Total cost / episode: $0.1008
• p50 / p95 latency: 1.47s / 4.82s
• JSON compliance: 0.55
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 35.1%
• Extraction methods: json_array_direct: 143, json_object_ads_key: 1, json_object_no_ads: 144, json_object_single_ad: 264, parse_failure: 300, regex_json_array: 3
• Verbosity: 1/855 calls over 1024 output tokens (0.1%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 378
• Extra keys observed: end_text, sponsor

mistralai/mistral-large-2512

• F1 (avg across episodes): 0.249
• Total cost / episode: $0.5599
• p50 / p95 latency: 2.55s / 6.20s
• JSON compliance: 1.00
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_array_direct: 855
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 1342
• Extra keys observed: end_text, sponsor

qwen/qwen3-14b

• F1 (avg across episodes): 0.248
• Total cost / episode: $0.1261
• p50 / p95 latency: 20.91s / 63.42s
• JSON compliance: 0.28
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 60.4%
• Extraction methods: json_object_no_ads: 1, json_object_single_ad: 338, parse_failure: 516
• Verbosity: 91/855 calls over 1024 output tokens (10.6%); 10 hit max_tokens (1.2%); 0 salvaged from truncated JSON (0.0%)

deepseek/deepseek-r1-distill-llama-70b

• F1 (avg across episodes): 0.241
• Total cost / episode: $0.7331
• p50 / p95 latency: 1.98s / 51.21s
• JSON compliance: 0.74
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 1.8%
• Extraction methods: json_array_direct: 20, json_object_ads_key: 68, json_object_no_ads: 104, json_object_single_ad: 642, json_object_single_ad_truncated: 5, parse_failure: 15, regex_json_array: 1
• Verbosity: 57/855 calls over 1024 output tokens (6.7%); 17 hit max_tokens (2.0%); 5 salvaged from truncated JSON (0.6%)
• Schema violations: 474
• Extra keys observed: end_text, sponsor

meta-llama/llama-4-scout

• F1 (avg across episodes): 0.235
• Total cost / episode: $0.0807
• p50 / p95 latency: 0.84s / 4.44s
• JSON compliance: 0.82
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 1.1%
• Extraction methods: bracket_fallback: 37, json_array_direct: 5, json_object_ads_key: 644, json_object_no_ads: 92, json_object_single_ad: 62, parse_failure: 9, regex_json_array: 6
• Verbosity: 1/855 calls over 1024 output tokens (0.1%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 507
• Extra keys observed: end_text, sponsor

nvidia/nemotron-nano-9b-v2

• F1 (avg across episodes): 0.221
• Total cost / episode: $0.0813
• p50 / p95 latency: 12.05s / 36.70s
• JSON compliance: 0.92
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 6.5%
• Extraction methods: json_array_direct: 770, json_object_single_ad_truncated: 15, parse_failure: 56, regex_json_array: 14
• Verbosity: 489/855 calls over 1024 output tokens (57.2%); 11 hit max_tokens (1.3%); 15 salvaged from truncated JSON (1.8%)
• Schema violations: 476
• Extra keys observed: end_text, sponsor

meta-llama/llama-3.1-8b-instruct

• F1 (avg across episodes): 0.198
• Total cost / episode: $0.0207
• p50 / p95 latency: 0.79s / 4.04s
• JSON compliance: 0.85
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.1%
• Extraction methods: json_array_direct: 371, json_object_no_ads: 66, json_object_single_ad: 417, parse_failure: 1
• Verbosity: 26/855 calls over 1024 output tokens (3.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 1270
• Extra keys observed: end_text, sponsor

cohere/command-r-plus-08-2024

• F1 (avg across episodes): 0.102
• Total cost / episode: $2.5787
• p50 / p95 latency: 0.95s / 3.45s
• JSON compliance: 0.98
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 0.0%
• Extraction methods: json_object_ads_key: 27, json_object_no_ads: 783, json_object_single_ad: 45
• Verbosity: 0/855 calls over 1024 output tokens (0.0%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 132
• Extra keys observed: end_text, sponsor

openai/o4-mini

• F1 (avg across episodes): 0.075
• Total cost / episode: $1.8999
• p50 / p95 latency: 6.95s / 25.84s
• JSON compliance: 0.05
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 93.9%
• Extraction methods: json_object_no_ads: 19, json_object_single_ad: 33, parse_failure: 803
• Verbosity: 340/855 calls over 1024 output tokens (39.8%); 12 hit max_tokens (1.4%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 28
• Extra keys observed: end_text, sponsor

microsoft/phi-4

• F1 (avg across episodes): 0.055
• Total cost / episode: $0.0713
• p50 / p95 latency: 2.29s / 11.55s
• JSON compliance: 0.86
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 1.1%
• Extraction methods: json_array_direct: 421, json_object_ads_key: 31, json_object_no_ads: 27, json_object_segments_key: 20, json_object_single_ad: 335, json_object_window_segments: 2, parse_failure: 9, regex_json_array: 10
• Verbosity: 19/855 calls over 1024 output tokens (2.2%); 12 hit max_tokens (1.4%); 0 salvaged from truncated JSON (0.0%)
• Schema violations: 412
• Extra keys observed: end_text, sponsor

qwen/qwen3-8b

• F1 (avg across episodes): 0.005
• Total cost / episode: $0.2578
• p50 / p95 latency: 59.45s / 144.22s
• JSON compliance: 0.01
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 97.4%
• Extraction methods: bracket_fallback: 18, json_array_direct: 4, parse_failure: 833
• Verbosity: 588/855 calls over 1024 output tokens (68.8%); 119 hit max_tokens (13.9%); 0 salvaged from truncated JSON (0.0%)

mistralai/mistral-7b-instruct-v0.1

• F1 (avg across episodes): 0.000
• Total cost / episode: $0.0285
• p50 / p95 latency: 7.11s / 33.05s
• JSON compliance: 0.16
• JSON mode: native (100% native, 855 calls)
• Parse failure rate: 59.5%
• Extraction methods: bracket_fallback: 1, parse_failure: 509, regex_json_array: 345
• Verbosity: 15/855 calls over 1024 output tokens (1.8%); 0 hit max_tokens (0.0%); 0 salvaged from truncated JSON (0.0%)

Per-Episode Detail

One subsection per episode in the corpus, showing how every model performed on that specific episode. For ad-bearing episodes you see F1 and the stdev across trials (low stdev means stable, high stdev means the model’s number on this episode is noisy). For the no-ad episode you see PASS / FAIL on the negative control: PASS = zero false positives across all windows, FAIL = the model flagged something that wasn’t an ad, with the count.

ep-ai-cloud-essentials-e8dc897fbd6b: How Physical AI is Streamlining Engineering

• Podcast: ai-cloud-essentials
• Duration: 16.4 min
• Truth: no-ads episode

Model	Result	FP count
openai/o4-mini	PASS	0
meta-llama/llama-3.1-8b-instruct	PASS	0
deepseek/deepseek-v4-pro	PASS	0
google/gemini-2.5-flash	PASS	0
qwen/qwen3-8b	PASS	0
claude-haiku-4-5-20251001	PASS	0
mistralai/mistral-large-2512	PASS	0
qwen/qwen3-14b	PASS	0
mistralai/codestral-2508	PASS	0
cohere/command-r-plus-08-2024	PASS	0
qwen/qwen3.6-plus	PASS	0
nvidia/llama-3.3-nemotron-super-49b-v1.5	PASS	0
qwen/qwen3.5-27b	PASS	0
meta-llama/llama-4-scout	PASS	0
meta-llama/llama-3.3-70b-instruct	PASS	0
mistralai/mistral-7b-instruct-v0.1	PASS	0
mistralai/mistral-medium-3.1	PASS	0
qwen/qwen3.5-plus-02-15	PASS	0
x-ai/grok-4.3	PASS	0
claude-sonnet-4-6	PASS	0
deepseek/deepseek-v3.2	PASS	0
claude-opus-4-7	PASS	0
openai/o3	PASS	0
google/gemini-3.5-flash	PASS	0
openai/gpt-oss-120b	FAIL	1
deepseek/deepseek-v4-flash	FAIL	1
deepseek/deepseek-r1	FAIL	1
qwen/qwen3.6-flash	FAIL	1
google/gemini-2.5-flash-lite	FAIL	1
google/gemini-3.1-flash-lite	FAIL	1
moonshotai/kimi-k2.6	FAIL	1
google/gemma-4-31b-it	FAIL	1
nvidia/nemotron-nano-9b-v2	FAIL	1
openai/gpt-5.4-mini	FAIL	1
google/gemini-2.5-pro	FAIL	1
meta-llama/llama-4-maverick	FAIL	1
openai/gpt-5.4	FAIL	1
openai/gpt-5.5	FAIL	1
qwen/qwen3-235b-a22b-2507	FAIL	2
deepseek/deepseek-r1-distill-llama-70b	FAIL	2
microsoft/phi-4	FAIL	3
openai/gpt-3.5-turbo	FAIL	3
cohere/command-a	FAIL	3
deepseek/deepseek-r1-0528	FAIL	27

ep-crime-junkie-8ce498f299d7: MISSING: Christopher “Cole” Thomas

• Podcast: crime-junkie
• Duration: 48.2 min
• Truth ads: 4

Model	F1	F1 stdev
qwen/qwen3.6-flash	1.000	0.000
qwen/qwen3.5-plus-02-15	1.000	0.000
meta-llama/llama-4-maverick	1.000	0.000
openai/gpt-3.5-turbo	0.978	0.050
google/gemini-2.5-pro	0.971	0.064
x-ai/grok-4.3	0.938	0.091
qwen/qwen3.6-plus	0.933	0.061
openai/gpt-5.4	0.933	0.061
google/gemini-3.5-flash	0.914	0.078
cohere/command-a	0.911	0.050
openai/gpt-5.5	0.898	0.059
qwen/qwen3.5-27b	0.892	0.062
claude-sonnet-4-6	0.889	0.000
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.888	0.115
openai/gpt-5.4-mini	0.883	0.089
google/gemma-4-31b-it	0.879	0.097
openai/o3	0.876	0.137
claude-opus-4-7	0.857	0.000
openai/gpt-oss-120b	0.819	0.152
qwen/qwen3-235b-a22b-2507	0.774	0.143
google/gemini-3.1-flash-lite	0.771	0.040
deepseek/deepseek-r1	0.732	0.156
deepseek/deepseek-v4-pro	0.728	0.201
deepseek/deepseek-v4-flash	0.718	0.106
deepseek/deepseek-r1-0528	0.717	0.187
deepseek/deepseek-v3.2	0.590	0.074
deepseek/deepseek-r1-distill-llama-70b	0.590	0.230
moonshotai/kimi-k2.6	0.547	0.203
mistralai/codestral-2508	0.520	0.038
meta-llama/llama-3.3-70b-instruct	0.502	0.183
claude-haiku-4-5-20251001	0.500	0.000
mistralai/mistral-large-2512	0.492	0.017
google/gemini-2.5-flash	0.462	0.000
google/gemini-2.5-flash-lite	0.462	0.000
mistralai/mistral-medium-3.1	0.397	0.115
nvidia/nemotron-nano-9b-v2	0.334	0.117
meta-llama/llama-4-scout	0.253	0.167
qwen/qwen3-14b	0.190	0.294
meta-llama/llama-3.1-8b-instruct	0.164	0.157
microsoft/phi-4	0.157	0.089
openai/o4-mini	0.000	0.000
qwen/qwen3-8b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-daily-gist-chicago-70a82fe93a5c: Suburban apartment market heats up

• Podcast: daily-gist-chicago
• Duration: 21.2 min
• Truth ads: 2

Model	F1	F1 stdev
qwen/qwen3-235b-a22b-2507	0.853	0.145
openai/gpt-5.4-mini	0.833	0.236
google/gemini-2.5-flash	0.800	0.000
claude-haiku-4-5-20251001	0.800	0.000
google/gemini-3.1-flash-lite	0.800	0.000
claude-sonnet-4-6	0.800	0.183
qwen/qwen3.5-27b	0.727	0.186
meta-llama/llama-3.1-8b-instruct	0.700	0.183
qwen/qwen3.6-flash	0.700	0.245
deepseek/deepseek-r1	0.693	0.213
mistralai/codestral-2508	0.667	0.000
qwen/qwen3.6-plus	0.667	0.000
google/gemini-2.5-flash-lite	0.667	0.000
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.667	0.000
mistralai/mistral-medium-3.1	0.667	0.000
mistralai/mistral-large-2512	0.648	0.043
deepseek/deepseek-r1-0528	0.647	0.228
openai/gpt-oss-120b	0.600	0.365
nvidia/nemotron-nano-9b-v2	0.600	0.365
qwen/qwen3.5-plus-02-15	0.600	0.091
deepseek/deepseek-v4-flash	0.587	0.335
meta-llama/llama-4-scout	0.587	0.335
x-ai/grok-4.3	0.533	0.075
openai/gpt-5.4	0.500	0.000
openai/gpt-5.5	0.500	0.000
google/gemma-4-31b-it	0.480	0.045
deepseek/deepseek-r1-distill-llama-70b	0.440	0.311
openai/gpt-3.5-turbo	0.420	0.045
google/gemini-2.5-pro	0.400	0.000
claude-opus-4-7	0.400	0.000
cohere/command-a	0.400	0.000
google/gemini-3.5-flash	0.400	0.000
deepseek/deepseek-v3.2	0.300	0.274
deepseek/deepseek-v4-pro	0.133	0.298
moonshotai/kimi-k2.6	0.100	0.224
meta-llama/llama-3.3-70b-instruct	0.100	0.224
openai/o4-mini	0.000	0.000
microsoft/phi-4	0.000	0.000
qwen/qwen3-8b	0.000	0.000
qwen/qwen3-14b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000
openai/o3	0.000	0.000
meta-llama/llama-4-maverick	0.000	0.000

ep-daily-tech-news-show-b576979e1fe8: Motorola Razr Fold is a Noble Competitor to the Galaxy Z Fold 7 - DTNS 5269

• Podcast: daily-tech-news-show
• Duration: 34.6 min
• Truth ads: 4

Model	F1	F1 stdev
openai/gpt-5.4-mini	0.943	0.078
qwen/qwen3.6-plus	0.914	0.078
moonshotai/kimi-k2.6	0.914	0.078
openai/gpt-5.4	0.892	0.062
openai/gpt-5.5	0.886	0.064
google/gemini-2.5-pro	0.864	0.089
claude-opus-4-7	0.863	0.014
qwen/qwen3.5-plus-02-15	0.857	0.000
google/gemini-3.5-flash	0.857	0.000
qwen/qwen3.6-flash	0.852	0.186
openai/o3	0.848	0.119
google/gemma-4-31b-it	0.811	0.132
meta-llama/llama-4-maverick	0.771	0.048
qwen/qwen3-235b-a22b-2507	0.756	0.265
qwen/qwen3.5-27b	0.743	0.104
deepseek/deepseek-r1-0528	0.700	0.245
deepseek/deepseek-r1	0.658	0.259
qwen/qwen3-14b	0.518	0.332
x-ai/grok-4.3	0.507	0.246
cohere/command-a	0.500	0.000
google/gemini-3.1-flash-lite	0.473	0.061
deepseek/deepseek-v4-flash	0.445	0.191
deepseek/deepseek-v4-pro	0.418	0.159
claude-sonnet-4-6	0.407	0.128
mistralai/codestral-2508	0.379	0.069
meta-llama/llama-3.3-70b-instruct	0.327	0.095
deepseek/deepseek-v3.2	0.327	0.211
google/gemini-2.5-flash-lite	0.326	0.036
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.299	0.206
nvidia/nemotron-nano-9b-v2	0.274	0.194
openai/gpt-oss-120b	0.270	0.117
google/gemini-2.5-flash	0.267	0.000
mistralai/mistral-large-2512	0.250	0.011
meta-llama/llama-4-scout	0.242	0.181
claude-haiku-4-5-20251001	0.235	0.000
meta-llama/llama-3.1-8b-instruct	0.231	0.132
openai/gpt-3.5-turbo	0.222	0.000
mistralai/mistral-medium-3.1	0.162	0.049
openai/o4-mini	0.147	0.202
deepseek/deepseek-r1-distill-llama-70b	0.057	0.128
microsoft/phi-4	0.000	0.000
qwen/qwen3-8b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-daily-tech-news-show-c1904b8605f7: Switch 2 Prices Rise, Forecast Drops - DTNS 5265

• Podcast: daily-tech-news-show
• Duration: 38.6 min
• Truth ads: 5

Model	F1	F1 stdev
moonshotai/kimi-k2.6	0.600	0.091
openai/gpt-5.5	0.547	0.064
qwen/qwen3.5-plus-02-15	0.518	0.025
openai/gpt-5.4	0.518	0.078
qwen/qwen3.5-27b	0.506	0.107
google/gemini-3.5-flash	0.500	0.000
qwen/qwen3.6-flash	0.492	0.017
openai/o3	0.472	0.150
openai/gpt-5.4-mini	0.468	0.144
qwen/qwen3.6-plus	0.457	0.036
google/gemini-2.5-pro	0.448	0.018
openai/gpt-3.5-turbo	0.364	0.000
deepseek/deepseek-v4-pro	0.350	0.229
cohere/command-r-plus-08-2024	0.313	0.155
deepseek/deepseek-v4-flash	0.310	0.043
cohere/command-a	0.298	0.090
claude-sonnet-4-6	0.237	0.037
claude-opus-4-7	0.231	0.082
deepseek/deepseek-r1-distill-llama-70b	0.205	0.017
meta-llama/llama-4-maverick	0.204	0.010
openai/gpt-oss-120b	0.201	0.141
deepseek/deepseek-r1-0528	0.184	0.066
x-ai/grok-4.3	0.179	0.092
mistralai/codestral-2508	0.172	0.046
deepseek/deepseek-r1	0.158	0.037
qwen/qwen3-235b-a22b-2507	0.142	0.087
deepseek/deepseek-v3.2	0.140	0.142
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.139	0.150
meta-llama/llama-4-scout	0.130	0.115
google/gemma-4-31b-it	0.119	0.058
qwen/qwen3-14b	0.100	0.224
mistralai/mistral-medium-3.1	0.095	0.010
google/gemini-3.1-flash-lite	0.076	0.001
mistralai/mistral-large-2512	0.074	0.000
claude-haiku-4-5-20251001	0.073	0.001
google/gemini-2.5-flash	0.071	0.000
google/gemini-2.5-flash-lite	0.069	0.040
nvidia/nemotron-nano-9b-v2	0.069	0.098
openai/o4-mini	0.067	0.149
microsoft/phi-4	0.056	0.077
meta-llama/llama-3.1-8b-instruct	0.029	0.064
qwen/qwen3-8b	0.000	0.000
meta-llama/llama-3.3-70b-instruct	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-drink-champs-30c9a2d49f13: Episode 501 w/ Warren Sapp

• Podcast: drink-champs
• Duration: 258.6 min
• Truth ads: 9

Model	F1	F1 stdev
openai/o3	0.742	0.113
qwen/qwen3.5-plus-02-15	0.711	0.022
qwen/qwen3.6-plus	0.706	0.027
google/gemini-2.5-pro	0.679	0.025
openai/gpt-5.5	0.642	0.125
qwen/qwen3.6-flash	0.628	0.076
openai/gpt-5.4	0.571	0.068
openai/gpt-5.4-mini	0.509	0.068
google/gemini-3.5-flash	0.496	0.021
qwen/qwen3.5-27b	0.494	0.184
claude-opus-4-7	0.334	0.070
claude-sonnet-4-6	0.275	0.085
meta-llama/llama-4-maverick	0.273	0.000
deepseek/deepseek-r1	0.268	0.153
deepseek/deepseek-v4-pro	0.262	0.111
deepseek/deepseek-v4-flash	0.218	0.063
x-ai/grok-4.3	0.208	0.047
deepseek/deepseek-v3.2	0.176	0.087
google/gemma-4-31b-it	0.158	0.035
deepseek/deepseek-r1-0528	0.125	0.058
google/gemini-2.5-flash-lite	0.084	0.006
microsoft/phi-4	0.058	0.053
openai/gpt-oss-120b	0.058	0.060
moonshotai/kimi-k2.6	0.053	0.074
mistralai/mistral-medium-3.1	0.046	0.003
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.045	0.067
qwen/qwen3-14b	0.045	0.061
nvidia/nemotron-nano-9b-v2	0.044	0.025
claude-haiku-4-5-20251001	0.042	0.000
openai/o4-mini	0.040	0.089
google/gemini-3.1-flash-lite	0.039	0.002
openai/gpt-3.5-turbo	0.038	0.035
google/gemini-2.5-flash	0.037	0.000
mistralai/mistral-large-2512	0.033	0.001
qwen/qwen3-235b-a22b-2507	0.033	0.031
mistralai/codestral-2508	0.033	0.030
cohere/command-a	0.031	0.043
meta-llama/llama-3.1-8b-instruct	0.017	0.015
qwen/qwen3-8b	0.000	0.000
deepseek/deepseek-r1-distill-llama-70b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
meta-llama/llama-4-scout	0.000	0.000
meta-llama/llama-3.3-70b-instruct	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-glt1412515089-373d5ba5007b: #2496 - Julia Mossbridge

• Podcast: glt1412515089
• Duration: 165.3 min
• Truth ads: 4

Model	F1	F1 stdev
google/gemini-3.5-flash	0.676	0.070
mistralai/mistral-medium-3.1	0.671	0.098
openai/o3	0.664	0.063
google/gemini-2.5-pro	0.646	0.028
google/gemma-4-31b-it	0.645	0.085
deepseek/deepseek-v4-flash	0.642	0.072
openai/gpt-5.5	0.636	0.143
deepseek/deepseek-r1	0.630	0.067
qwen/qwen3.5-plus-02-15	0.625	0.057
openai/gpt-oss-120b	0.625	0.084
qwen/qwen3.5-27b	0.601	0.116
qwen/qwen3.6-plus	0.600	0.000
claude-opus-4-7	0.600	0.000
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.596	0.074
x-ai/grok-4.3	0.572	0.131
claude-haiku-4-5-20251001	0.571	0.000
meta-llama/llama-3.3-70b-instruct	0.567	0.030
qwen/qwen3-235b-a22b-2507	0.540	0.070
google/gemini-3.1-flash-lite	0.527	0.015
qwen/qwen3.6-flash	0.518	0.025
deepseek/deepseek-v3.2	0.518	0.085
cohere/command-a	0.507	0.027
meta-llama/llama-4-maverick	0.507	0.015
openai/gpt-5.4	0.506	0.059
google/gemini-2.5-flash	0.500	0.000
openai/gpt-5.4-mini	0.494	0.052
moonshotai/kimi-k2.6	0.469	0.066
deepseek/deepseek-v4-pro	0.449	0.128
qwen/qwen3-14b	0.434	0.153
google/gemini-2.5-flash-lite	0.413	0.020
claude-sonnet-4-6	0.400	0.000
deepseek/deepseek-r1-0528	0.379	0.061
meta-llama/llama-4-scout	0.336	0.138
deepseek/deepseek-r1-distill-llama-70b	0.265	0.036
openai/gpt-3.5-turbo	0.254	0.006
mistralai/mistral-large-2512	0.253	0.039
nvidia/nemotron-nano-9b-v2	0.242	0.087
mistralai/codestral-2508	0.231	0.091
meta-llama/llama-3.1-8b-instruct	0.183	0.109
openai/o4-mini	0.080	0.179
microsoft/phi-4	0.000	0.000
qwen/qwen3-8b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-it-s-a-thing-e339179dfad6: SOUP shots - It’s a Thing 418

• Podcast: it-s-a-thing
• Duration: 26.7 min
• Truth ads: 1

Model	F1	F1 stdev
qwen/qwen3.6-flash	0.667	0.000
qwen/qwen3.6-plus	0.667	0.000
google/gemini-2.5-pro	0.667	0.000
qwen/qwen3.5-plus-02-15	0.667	0.000
claude-opus-4-7	0.667	0.000
google/gemini-3.5-flash	0.667	0.000
openai/gpt-5.5	0.667	0.000
openai/gpt-5.4	0.613	0.119
qwen/qwen3.5-27b	0.600	0.365
openai/gpt-5.4-mini	0.500	0.289
google/gemma-4-31b-it	0.467	0.274
google/gemini-3.1-flash-lite	0.433	0.253
x-ai/grok-4.3	0.433	0.253
deepseek/deepseek-r1-0528	0.404	0.281
meta-llama/llama-3.1-8b-instruct	0.400	0.548
deepseek/deepseek-v3.2	0.400	0.548
cohere/command-a	0.400	0.000
deepseek/deepseek-v4-flash	0.337	0.239
openai/o3	0.333	0.471
deepseek/deepseek-r1	0.313	0.301
deepseek/deepseek-v4-pro	0.267	0.365
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.233	0.325
openai/gpt-oss-120b	0.227	0.209
moonshotai/kimi-k2.6	0.200	0.274
openai/o4-mini	0.000	0.000
microsoft/phi-4	0.000	0.000
qwen/qwen3-235b-a22b-2507	0.000	0.000
google/gemini-2.5-flash	0.000	0.000
qwen/qwen3-8b	0.000	0.000
claude-haiku-4-5-20251001	0.000	0.000
mistralai/mistral-large-2512	0.000	0.000
deepseek/deepseek-r1-distill-llama-70b	0.000	0.000
qwen/qwen3-14b	0.000	0.000
mistralai/codestral-2508	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
openai/gpt-3.5-turbo	0.000	0.000
google/gemini-2.5-flash-lite	0.000	0.000
meta-llama/llama-4-scout	0.000	0.000
meta-llama/llama-3.3-70b-instruct	0.000	0.000
nvidia/nemotron-nano-9b-v2	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000
mistralai/mistral-medium-3.1	0.000	0.000
claude-sonnet-4-6	0.000	0.000
meta-llama/llama-4-maverick	0.000	0.000

ep-on-air-with-dan-and-alex2-574e4f303730: Ryanair Wants Alcohol Bans, Emirates’ $6.8B Record Profit & Buying Spirit Airlines?!

• Podcast: on-air-with-dan-and-alex2
• Duration: 58.1 min
• Truth ads: 2

Model	F1	F1 stdev
qwen/qwen3.6-plus	0.773	0.060
openai/gpt-5.4	0.747	0.073
moonshotai/kimi-k2.6	0.734	0.200
openai/o3	0.687	0.124
qwen/qwen3.6-flash	0.682	0.115
qwen/qwen3.5-27b	0.634	0.194
openai/gpt-5.4-mini	0.610	0.052
claude-opus-4-7	0.595	0.071
qwen/qwen3.5-plus-02-15	0.590	0.043
meta-llama/llama-4-maverick	0.571	0.000
google/gemini-3.5-flash	0.571	0.000
openai/gpt-5.5	0.571	0.000
deepseek/deepseek-r1	0.555	0.176
google/gemini-2.5-pro	0.543	0.039
deepseek/deepseek-v4-flash	0.535	0.112
qwen/qwen3-235b-a22b-2507	0.517	0.054
google/gemma-4-31b-it	0.514	0.032
deepseek/deepseek-r1-0528	0.514	0.122
cohere/command-a	0.503	0.045
claude-haiku-4-5-20251001	0.500	0.000
mistralai/mistral-medium-3.1	0.500	0.000
x-ai/grok-4.3	0.472	0.066
mistralai/codestral-2508	0.469	0.045
openai/gpt-oss-120b	0.457	0.178
google/gemini-2.5-flash	0.444	0.000
mistralai/mistral-large-2512	0.444	0.000
claude-sonnet-4-6	0.444	0.000
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.431	0.109
openai/gpt-3.5-turbo	0.400	0.000
google/gemini-2.5-flash-lite	0.400	0.000
google/gemini-3.1-flash-lite	0.400	0.000
nvidia/nemotron-nano-9b-v2	0.361	0.091
deepseek/deepseek-r1-distill-llama-70b	0.343	0.062
qwen/qwen3-14b	0.330	0.242
deepseek/deepseek-v3.2	0.300	0.274
meta-llama/llama-3.1-8b-instruct	0.284	0.166
deepseek/deepseek-v4-pro	0.257	0.433
microsoft/phi-4	0.213	0.307
openai/o4-mini	0.133	0.298
meta-llama/llama-3.3-70b-instruct	0.133	0.298
meta-llama/llama-4-scout	0.094	0.130
qwen/qwen3-8b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-oxide-and-friends-ce789ff5b62e: Mechanical Engineering at Oxide [chapter images]

• Podcast: oxide-and-friends
• Duration: 84.5 min
• Truth: no-ads episode

Model	Result	FP count
openai/o4-mini	PASS	0
meta-llama/llama-3.1-8b-instruct	PASS	0
deepseek/deepseek-v4-pro	PASS	0
google/gemini-2.5-flash	PASS	0
qwen/qwen3-8b	PASS	0
openai/gpt-oss-120b	PASS	0
deepseek/deepseek-v4-flash	PASS	0
claude-haiku-4-5-20251001	PASS	0
mistralai/mistral-large-2512	PASS	0
qwen/qwen3.6-flash	PASS	0
mistralai/codestral-2508	PASS	0
cohere/command-r-plus-08-2024	PASS	0
qwen/qwen3.6-plus	PASS	0
google/gemini-2.5-flash-lite	PASS	0
google/gemini-3.1-flash-lite	PASS	0
nvidia/llama-3.3-nemotron-super-49b-v1.5	PASS	0
qwen/qwen3.5-27b	PASS	0
meta-llama/llama-4-scout	PASS	0
meta-llama/llama-3.3-70b-instruct	PASS	0
google/gemma-4-31b-it	PASS	0
nvidia/nemotron-nano-9b-v2	PASS	0
mistralai/mistral-7b-instruct-v0.1	PASS	0
mistralai/mistral-medium-3.1	PASS	0
qwen/qwen3.5-plus-02-15	PASS	0
x-ai/grok-4.3	PASS	0
claude-sonnet-4-6	PASS	0
claude-opus-4-7	PASS	0
openai/o3	PASS	0
cohere/command-a	PASS	0
meta-llama/llama-4-maverick	PASS	0
google/gemini-3.5-flash	PASS	0
openai/gpt-5.5	PASS	0
deepseek/deepseek-r1	FAIL	1
qwen/qwen3-14b	FAIL	1
openai/gpt-5.4-mini	FAIL	1
google/gemini-2.5-pro	FAIL	1
openai/gpt-5.4	FAIL	1
deepseek/deepseek-v3.2	FAIL	2
moonshotai/kimi-k2.6	FAIL	4
qwen/qwen3-235b-a22b-2507	FAIL	6
deepseek/deepseek-r1-distill-llama-70b	FAIL	10
openai/gpt-3.5-turbo	FAIL	10
deepseek/deepseek-r1-0528	FAIL	12
microsoft/phi-4	FAIL	15

ep-security-now-audio-2850b24903b2: SN 1077: A Browser AI API? - End of Bug Bounties?

• Podcast: security-now-audio
• Duration: 156.2 min
• Truth ads: 6

Model	F1	F1 stdev
openai/o3	0.644	0.161
mistralai/mistral-medium-3.1	0.640	0.137
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.587	0.064
cohere/command-r-plus-08-2024	0.569	0.053
claude-haiku-4-5-20251001	0.551	0.012
deepseek/deepseek-v3.2	0.528	0.073
claude-opus-4-7	0.520	0.055
meta-llama/llama-4-scout	0.520	0.050
claude-sonnet-4-6	0.516	0.014
qwen/qwen3.6-flash	0.514	0.052
meta-llama/llama-3.3-70b-instruct	0.512	0.035
openai/gpt-5.5	0.505	0.047
deepseek/deepseek-v4-pro	0.498	0.089
google/gemma-4-31b-it	0.496	0.021
meta-llama/llama-4-maverick	0.496	0.021
openai/gpt-5.4	0.495	0.022
openai/gpt-5.4-mini	0.491	0.027
x-ai/grok-4.3	0.486	0.013
qwen/qwen3.6-plus	0.481	0.011
qwen/qwen3.5-plus-02-15	0.476	0.000
google/gemini-3.5-flash	0.476	0.000
openai/gpt-oss-120b	0.469	0.033
deepseek/deepseek-v4-flash	0.465	0.046
deepseek/deepseek-r1	0.462	0.044
qwen/qwen3-235b-a22b-2507	0.460	0.031
google/gemini-2.5-flash	0.455	0.000
google/gemini-2.5-pro	0.451	0.009
qwen/qwen3.5-27b	0.414	0.081
google/gemini-3.1-flash-lite	0.413	0.007
mistralai/codestral-2508	0.374	0.032
cohere/command-a	0.368	0.015
qwen/qwen3-14b	0.336	0.241
deepseek/deepseek-r1-0528	0.281	0.103
google/gemini-2.5-flash-lite	0.280	0.051
deepseek/deepseek-r1-distill-llama-70b	0.262	0.018
nvidia/nemotron-nano-9b-v2	0.238	0.144
openai/gpt-3.5-turbo	0.217	0.004
mistralai/mistral-large-2512	0.209	0.017
moonshotai/kimi-k2.6	0.196	0.245
meta-llama/llama-3.1-8b-instruct	0.133	0.052
openai/o4-mini	0.114	0.156
microsoft/phi-4	0.033	0.045
qwen/qwen3-8b	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-the-brilliant-idiots-0bb9bf634c8e: Class Rank

• Podcast: the-brilliant-idiots
• Duration: 119.9 min
• Truth ads: 3

Model	F1	F1 stdev
google/gemini-3.5-flash	0.857	0.000
claude-sonnet-4-6	0.836	0.048
qwen/qwen3.6-flash	0.800	0.128
qwen/qwen3.6-plus	0.798	0.087
openai/gpt-5.5	0.776	0.081
qwen/qwen3.5-plus-02-15	0.771	0.048
x-ai/grok-4.3	0.771	0.048
claude-opus-4-7	0.733	0.037
openai/o3	0.698	0.273
qwen/qwen3.5-27b	0.671	0.267
openai/gpt-oss-120b	0.630	0.067
claude-haiku-4-5-20251001	0.600	0.000
mistralai/mistral-medium-3.1	0.591	0.046
google/gemma-4-31b-it	0.585	0.077
deepseek/deepseek-v4-flash	0.569	0.045
moonshotai/kimi-k2.6	0.538	0.263
openai/gpt-5.4	0.516	0.110
google/gemini-2.5-pro	0.510	0.036
meta-llama/llama-3.3-70b-instruct	0.489	0.025
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.478	0.160
google/gemini-2.5-flash-lite	0.432	0.046
deepseek/deepseek-r1	0.401	0.043
meta-llama/llama-4-maverick	0.390	0.101
qwen/qwen3-14b	0.372	0.099
openai/gpt-5.4-mini	0.353	0.145
meta-llama/llama-4-scout	0.349	0.080
qwen/qwen3-235b-a22b-2507	0.342	0.064
google/gemini-3.1-flash-lite	0.338	0.072
cohere/command-a	0.247	0.013
deepseek/deepseek-r1-distill-llama-70b	0.222	0.056
openai/gpt-3.5-turbo	0.220	0.005
nvidia/nemotron-nano-9b-v2	0.205	0.056
mistralai/mistral-large-2512	0.193	0.041
mistralai/codestral-2508	0.187	0.024
openai/o4-mini	0.180	0.249
deepseek/deepseek-v4-pro	0.167	0.236
deepseek/deepseek-r1-0528	0.161	0.084
google/gemini-2.5-flash	0.125	0.000
deepseek/deepseek-v3.2	0.100	0.224
microsoft/phi-4	0.067	0.092
meta-llama/llama-3.1-8b-instruct	0.000	0.000
qwen/qwen3-8b	0.000	0.000
cohere/command-r-plus-08-2024	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-the-tim-dillon-show-f62bd5fa1cfe: 495 - Hantavirus Cruise & iPad Babies

• Podcast: the-tim-dillon-show
• Duration: 80.1 min
• Truth ads: 6

Model	F1	F1 stdev
qwen/qwen3.5-plus-02-15	0.636	0.028
qwen/qwen3.6-plus	0.615	0.000
claude-opus-4-7	0.592	0.052
google/gemini-3.5-flash	0.592	0.052
openai/gpt-5.4	0.586	0.077
google/gemini-2.5-pro	0.569	0.063
deepseek/deepseek-v4-pro	0.565	0.091
openai/gpt-5.5	0.546	0.063
qwen/qwen3.5-27b	0.537	0.134
qwen/qwen3.6-flash	0.507	0.183
qwen/qwen3-235b-a22b-2507	0.421	0.048
x-ai/grok-4.3	0.393	0.123
openai/gpt-oss-120b	0.387	0.066
openai/gpt-5.4-mini	0.373	0.103
openai/o3	0.372	0.081
google/gemma-4-31b-it	0.350	0.072
deepseek/deepseek-r1	0.327	0.195
meta-llama/llama-3.3-70b-instruct	0.308	0.000
deepseek/deepseek-v4-flash	0.298	0.084
deepseek/deepseek-r1-0528	0.283	0.115
qwen/qwen3-14b	0.233	0.158
mistralai/mistral-medium-3.1	0.223	0.054
cohere/command-a	0.200	0.078
moonshotai/kimi-k2.6	0.184	0.105
google/gemini-3.1-flash-lite	0.183	0.016
claude-sonnet-4-6	0.179	0.004
mistralai/codestral-2508	0.178	0.065
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.171	0.125
meta-llama/llama-4-maverick	0.167	0.000
deepseek/deepseek-r1-distill-llama-70b	0.163	0.159
google/gemini-2.5-flash	0.154	0.000
claude-haiku-4-5-20251001	0.154	0.000
google/gemini-2.5-flash-lite	0.133	0.026
openai/gpt-3.5-turbo	0.125	0.000
microsoft/phi-4	0.079	0.072
deepseek/deepseek-v3.2	0.057	0.079
qwen/qwen3-8b	0.057	0.128
cohere/command-r-plus-08-2024	0.057	0.128
meta-llama/llama-4-scout	0.053	0.049
meta-llama/llama-3.1-8b-instruct	0.049	0.019
nvidia/nemotron-nano-9b-v2	0.044	0.061
mistralai/mistral-large-2512	0.044	0.025
openai/o4-mini	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

ep-tosh-show-5f6894439bb6: My Mom - Emergency Pod

• Podcast: tosh-show
• Duration: 41.4 min
• Truth ads: 5

Model	F1	F1 stdev
qwen/qwen3.6-plus	0.703	0.119
qwen/qwen3.5-plus-02-15	0.696	0.094
qwen/qwen3.5-27b	0.647	0.125
deepseek/deepseek-r1-0528	0.643	0.158
google/gemini-2.5-pro	0.603	0.116
openai/gpt-5.5	0.587	0.084
x-ai/grok-4.3	0.585	0.075
deepseek/deepseek-v4-flash	0.574	0.137
openai/gpt-5.4	0.569	0.070
mistralai/codestral-2508	0.564	0.149
qwen/qwen3.6-flash	0.564	0.017
openai/gpt-5.4-mini	0.547	0.064
cohere/command-a	0.533	0.122
claude-opus-4-7	0.524	0.135
google/gemini-3.5-flash	0.490	0.039
openai/o3	0.464	0.106
openai/gpt-oss-120b	0.443	0.185
deepseek/deepseek-r1	0.435	0.066
mistralai/mistral-medium-3.1	0.421	0.048
qwen/qwen3-14b	0.412	0.193
meta-llama/llama-4-maverick	0.400	0.000
nvidia/llama-3.3-nemotron-super-49b-v1.5	0.399	0.223
claude-haiku-4-5-20251001	0.375	0.000
google/gemini-3.1-flash-lite	0.375	0.000
claude-sonnet-4-6	0.375	0.000
google/gemini-2.5-flash-lite	0.364	0.052
qwen/qwen3-235b-a22b-2507	0.362	0.048
mistralai/mistral-large-2512	0.353	0.000
meta-llama/llama-3.3-70b-instruct	0.347	0.087
deepseek/deepseek-r1-distill-llama-70b	0.340	0.152
deepseek/deepseek-v4-pro	0.305	0.312
cohere/command-r-plus-08-2024	0.289	0.087
moonshotai/kimi-k2.6	0.267	0.149
meta-llama/llama-4-scout	0.258	0.066
nvidia/nemotron-nano-9b-v2	0.240	0.125
google/gemini-2.5-flash	0.200	0.000
openai/gpt-3.5-turbo	0.189	0.010
meta-llama/llama-3.1-8b-instruct	0.185	0.027
google/gemma-4-31b-it	0.182	0.025
deepseek/deepseek-v3.2	0.140	0.219
openai/o4-mini	0.133	0.183
microsoft/phi-4	0.000	0.000
qwen/qwen3-8b	0.000	0.000
mistralai/mistral-7b-instruct-v0.1	0.000	0.000

Parser stress test

How each model’s responses were actually parsed. Columns are extraction methods, ordered alphabetically; rows are models, sorted by parse-failure rate (cleanest at top). json_array_direct is the happy path: a bare JSON array we could json.loads and process immediately. markdown_code_block means we had to strip triple-backtick fences first; json_object_* means the model wrapped the array in an outer object and we had to find the array key; regex_* are last-resort recovery paths. A model that needs anything but json_array_direct for most calls is fragile. It works today, but a small prompt change can break the parser.

Model	bracket_fallback	json_array_direct	json_object_ads_key	json_object_no_ads	json_object_segments_key	json_object_single_ad	json_object_single_ad_truncated	json_object_window_segments	markdown_code_block	parse_failure	regex_json_array
qwen/qwen3-235b-a22b-2507	0	173	1	94	0	587	0	0	0	0	0
google/gemini-2.5-flash	0	855	0	0	0	0	0	0	0	0	0
claude-haiku-4-5-20251001	0	0	0	0	0	0	0	0	855	0	0
mistralai/mistral-large-2512	0	855	0	0	0	0	0	0	0	0	0
qwen/qwen3.6-flash	0	855	0	0	0	0	0	0	0	0	0
mistralai/codestral-2508	0	855	0	0	0	0	0	0	0	0	0
cohere/command-r-plus-08-2024	0	0	27	783	0	45	0	0	0	0	0
qwen/qwen3.6-plus	0	855	0	0	0	0	0	0	0	0	0
google/gemini-3.1-flash-lite	0	799	0	0	0	0	0	0	0	0	56
mistralai/mistral-medium-3.1	0	854	0	0	0	0	1	0	0	0	0
openai/gpt-5.4-mini	0	0	2	300	2	551	0	0	0	0	0
claude-sonnet-4-6	0	783	0	0	0	0	0	0	57	0	15
deepseek/deepseek-v3.2	0	501	16	3	0	335	0	0	0	0	0
claude-opus-4-7	0	850	0	0	0	0	0	0	0	0	5
cohere/command-a	0	0	0	29	0	826	0	0	0	0	0
meta-llama/llama-4-maverick	0	0	0	315	0	540	0	0	0	0	0
google/gemini-3.5-flash	0	850	0	0	0	0	1	0	0	0	4
openai/gpt-5.4	0	0	0	326	0	529	0	0	0	0	0
meta-llama/llama-3.1-8b-instruct	0	371	0	66	0	417	0	0	0	1	0
google/gemma-4-31b-it	1	0	455	216	0	180	2	0	0	1	0
x-ai/grok-4.3	0	854	0	0	0	0	0	0	0	1	0
openai/gpt-5.5	0	0	0	494	0	360	0	0	0	1	0
qwen/qwen3.5-plus-02-15	0	851	0	0	0	0	0	0	0	1	0
openai/gpt-3.5-turbo	0	0	0	50	0	800	0	0	0	5	0
openai/o3	0	0	34	621	12	183	0	0	0	5	0
deepseek/deepseek-r1	0	760	2	17	7	44	0	0	17	6	2
microsoft/phi-4	0	421	31	27	20	335	0	2	0	9	10
meta-llama/llama-4-scout	37	5	644	92	0	62	0	0	0	9	6
google/gemini-2.5-pro	0	818	0	0	0	0	0	0	0	12	25
deepseek/deepseek-r1-distill-llama-70b	0	20	68	104	0	642	5	0	0	15	1
deepseek/deepseek-v4-flash	0	102	435	21	7	272	0	0	0	17	1
google/gemini-2.5-flash-lite	0	793	0	0	0	0	43	0	0	19	0
deepseek/deepseek-v4-pro	0	229	63	131	314	86	2	0	5	23	2
nvidia/nemotron-nano-9b-v2	0	770	0	0	0	0	15	0	0	56	14
deepseek/deepseek-r1-0528	0	700	34	3	0	30	3	0	2	83	0
qwen/qwen3.5-27b	0	633	0	89	0	12	0	0	0	121	0
nvidia/llama-3.3-nemotron-super-49b-v1.5	0	441	0	0	0	0	3	0	257	124	30
openai/gpt-oss-120b	0	68	235	186	0	188	6	0	0	149	23
moonshotai/kimi-k2.6	0	68	35	109	2	397	0	0	5	239	0
meta-llama/llama-3.3-70b-instruct	0	143	1	144	0	264	0	0	0	300	3
mistralai/mistral-7b-instruct-v0.1	1	0	0	0	0	0	0	0	0	509	345
qwen/qwen3-14b	0	0	0	1	0	338	0	0	0	516	0
openai/o4-mini	0	0	0	19	0	33	0	0	0	803	0
qwen/qwen3-8b	18	4	0	0	0	0	0	0	0	833	0

Methodology

Reproducibility settings used for this run. The benchmark sends the same prompts MinusPod sends in production (same system prompt, same sponsor list, same windowing) so the F1 numbers here are directly relevant to production accuracy decisions. Cost is recomputed at report time from token counts against the active pricing snapshot, so all rows compare at the same prices regardless of when the actual call ran.

• Trials per (model, episode): 5, temperature 0.0
• max_tokens: 4096 (matches MinusPod production)
• response_format: json_object (with prompt-injection fallback when provider rejects native)
• Window size: 10 min, overlap: 3 min (imported from MinusPod’s create_windows)
• Pricing snapshot: 2026-05-20T19:59:56.229629Z
• Corpus episodes: 14

Transcript source

segments.json for every corpus episode is pulled byte-exact from the source MinusPod instance’s original-segments endpoint. The transcript itself was generated by faster-whisper inside that instance, not by the benchmark. Model choice and decoding params affect what gets transcribed, which sets an upper bound on what every benchmarked LLM can find.

Whisper config:

Setting	Value
Model	large-v3
Backend	local (faster-whisper, CUDA GPU)
Compute type	auto (resolves to float16 on CUDA)
Language	en (forced English, not auto-detect)
VAD gap detection	on (start 3.0s / mid 8.0s / tail 3.0s)

model.transcribe() invocation (from src/transcriber.py):

WhisperModel(model_size="large-v3", device="cuda", compute_type="auto")
model.transcribe(
    audio,
    language="en",
    initial_prompt=<podcast name + SEED_SPONSORS vocabulary>,
    beam_size=5,
    batch_size=<adaptive: 16/12/8/4 by episode length>,
    word_timestamps=True,
    vad_filter=True,
    vad_parameters={"min_silence_duration_ms": 1000, "speech_pad_ms": 600, "threshold": 0.3},
)

The initial_prompt carries a sponsor vocabulary so Whisper produces consistent spellings (Athletic Greens rather than AG1, ExpressVPN rather than express vpn). This biases what shows up in the transcript and therefore what every benchmarked LLM is scored against.

Sponsor vocabulary (254 canonical sponsors, 43 of them with explicit alias spellings totaling 46 aliases; from src/utils/constants.py SEED_SPONSORS). Laid out in two side-by-side groups, read top-to-bottom in each group.

Sponsor	Aliases	Category	Sponsor	Aliases	Category
1Password	One Password	tech	Lyft	-	automotive_transport
Acorns	-	finance	MacPaw	CleanMyMac	tech
ADT	-	home	Magic Mind	-	beverage
Affirm	-	finance_fintech	Magic Spoon	-	food
Airbnb	-	travel_hospitality	Mailchimp	-	tech_software_saas
Airtable	-	tech_software_saas	Manscaped	-	personal
Alani Nu	-	food_beverage_nutrition	MasterClass	Master Class	education
Allbirds	-	ecommerce_retail_dtc	McDonald’s	-	food_beverage_nutrition
Alo Yoga	-	ecommerce_retail_dtc	Mercury	-	finance_fintech
Amazon	-	retail	Meter	-	b2b_startup
Anthropic	-	tech_software_saas	Midjourney	-	tech_software_saas
Apple TV+	-	media_streaming	Mint Mobile	MintMobile	telecom
Asana	-	tech_software_saas	Miro	-	tech
AT&T	-	telecom	Momentous	-	mental_health_wellness
Athletic Brewing	-	beverage	Monarch Money	-	finance
Athletic Greens	AG1, AG One	health	Monday.com	Monday	tech
Audible	-	entertainment	Native	-	personal
Aura	-	tech	NerdWallet	-	finance_fintech
Babbel	-	education	Netflix	-	media_streaming
BetMGM	Bet MGM	gambling	NetSuite	Net Suite	tech
BetterHelp	Better Help	health	Noom	-	mental_health_wellness
Betterment	-	finance	NordVPN	Nord VPN	vpn
Bill.com	-	finance_fintech	Notion	-	tech
Birchbox	-	ecommerce_retail_dtc	Nutrafol	-	health
Bitwarden	Bit Warden	tech	Okta	-	tech_software_saas
Blinkist	-	education	OLIPOP	-	food_beverage_nutrition
Bloom Nutrition	-	food_beverage_nutrition	OneSkin	One Skin	personal
Blue Apron	-	food	OpenAI	-	tech_software_saas
Bombas	-	apparel	Outdoor Voices	-	ecommerce_retail_dtc
Booking.com	-	travel_hospitality	OutSystems	-	tech
Bose	-	electronics	PagerDuty	-	b2b_startup
Brex	-	finance_fintech	Paramount+	-	media_streaming
Brilliant	-	tech_software_saas	Patreon	-	tech_software_saas
Brooklinen	-	home	Perplexity	-	tech_software_saas
Butcher Box	ButcherBox	food	Plaid	-	finance_fintech
CacheFly	-	tech	PolicyGenius	Policy Genius	finance
Caesars Sportsbook	-	gaming_sports_betting	Poppi	-	food_beverage_nutrition
Calm	-	health	Poshmark	-	ecommerce_retail_dtc
Canva	-	tech	Progressive	-	finance
Capital One	-	finance	Public.com	-	finance_fintech
Care/of	Care of, Careof	health	Pura	-	home_security
CarMax	Car Max	auto	Purple	-	home
Carvana	-	auto	QuickBooks	-	finance_fintech
Casper	-	home	Quince	-	apparel
Cerebral	-	mental_health_wellness	Quip	-	personal
Chime	-	finance_fintech	Ramp	-	finance_fintech
ClickUp	-	tech_software_saas	Raycon	-	electronics
Cloudflare	-	tech_software_saas	Retool	-	tech_software_saas
Coinbase	-	finance_fintech	Ring	-	home
Comcast	-	telecom	Rippling	-	b2b_startup
Cozy Earth	-	home	Ritual	-	health
Credit Karma	-	finance	Ro	-	mental_health_wellness
CrowdStrike	-	tech_software_saas	Robinhood	-	finance_fintech
Cursor	-	tech_software_saas	Rocket Lawyer	-	insurance_legal
Databricks	-	tech_software_saas	Rocket Money	RocketMoney, Truebill	finance
Datadog	-	tech_software_saas	Roman	-	health
Deel	-	business	Rosetta Stone	-	education
DeleteMe	Delete Me	tech	Rothy’s	-	ecommerce_retail_dtc
Disney+	-	media_streaming	Saatva	-	ecommerce_retail_dtc
DocuSign	-	tech_software_saas	Salesforce	-	tech_software_saas
Dollar Shave Club	DSC	personal	SeatGeek	-	gaming_sports_betting
DoorDash	Door Dash	food	Seed	-	health
DraftKings	Draft Kings	gambling	SendGrid	-	tech_software_saas
Duolingo	-	tech_software_saas	ServiceNow	-	tech_software_saas
eBay Motors	-	auto	Shein	-	ecommerce_retail_dtc
Eight Sleep	-	mental_health_wellness	Shopify	-	tech
ElevenLabs	-	tech_software_saas	SimpliSafe	Simpli Safe	home
ESPN Bet	-	gaming_sports_betting	SiriusXM	-	media_streaming
Everlane	-	ecommerce_retail_dtc	Skillshare	-	tech_software_saas
EveryPlate	-	food_beverage_nutrition	SKIMS	-	ecommerce_retail_dtc
Expedia	-	travel_hospitality	Skyscanner	-	travel_hospitality
ExpressVPN	Express VPN	vpn	Slack	-	tech_software_saas
FabFitFun	-	ecommerce_retail_dtc	Snowflake	-	tech_software_saas
Factor	-	food	SoFi	-	finance
FanDuel	Fan Duel	gambling	Spaceship	-	tech
Figma	-	tech_software_saas	Splunk	-	b2b_startup
Ford	-	auto	Spotify	-	media_streaming
Framer	-	tech	Squarespace	Square Space	tech
FreshBooks	-	finance_fintech	Stamps.com	Stamps	business
Function Health	-	mental_health_wellness	Starbucks	-	food_beverage_nutrition
Function of Beauty	-	personal	State Farm	-	finance
Gametime	Game Time	entertainment	Stitch Fix	-	ecommerce_retail_dtc
Geico	-	finance	StockX	-	ecommerce_retail_dtc
GitHub	-	tech_software_saas	Stripe	-	finance_fintech
GitHub Copilot	-	tech_software_saas	StubHub	-	gaming_sports_betting
GOAT	-	ecommerce_retail_dtc	Substack	-	tech_software_saas
GoodRx	Good Rx	health	T-Mobile	TMobile	telecom
Gopuff	-	ecommerce_retail_dtc	Talkspace	-	mental_health_wellness
Grammarly	-	tech	Temu	-	ecommerce_retail_dtc
Green Chef	GreenChef	food	Ten Thousand	-	ecommerce_retail_dtc
Grubhub	Grub Hub	food	Thinkst Canary	-	tech
Gusto	-	b2b_startup	Thorne	-	mental_health_wellness
Harry’s	Harrys	personal	ThreatLocker	-	tech
HBO Max	-	media_streaming	ThredUp	-	ecommerce_retail_dtc
Headspace	Head Space	health	Thrive Market	-	food
Helix Sleep	Helix	home	Toyota	-	auto
HelloFresh	Hello Fresh	food	Transparent Labs	-	food_beverage_nutrition
Hers	-	health	Turo	-	automotive_transport
Hims	-	health	Twilio	-	tech_software_saas
Honeylove	Honey Love	apparel	Uber	-	automotive_transport
Hopper	-	travel_hospitality	Uber Eats	UberEats	food
HubSpot	Hub Spot	tech	UnitedHealth Group	-	finance_fintech
Huel	-	food_beverage_nutrition	Vanta	-	tech
Hyundai	-	auto	Veeam	-	tech
iHeartRadio	-	media_streaming	Vercel	-	tech_software_saas
Imperfect Foods	-	food_beverage_nutrition	Verizon	-	telecom
Incogni	-	tech	Visible	-	telecom
Indeed	-	jobs	Vrbo	-	travel_hospitality
Inside Tracker	-	mental_health_wellness	Vuori	-	ecommerce_retail_dtc
Instacart	-	food	Warby Parker	-	ecommerce_retail_dtc
Intuit	-	finance_fintech	Wayfair	-	ecommerce_retail_dtc
Joovv	-	mental_health_wellness	Waymo	-	automotive_transport
Kayak	-	travel_hospitality	Wealthfront	-	finance
Keeps	-	health	WebBank	-	finance_fintech
Klarna	-	finance_fintech	Webflow	-	b2b_startup
Klaviyo	-	tech_software_saas	WhatsApp	-	tech
LegalZoom	-	insurance_legal	WHOOP	-	mental_health_wellness
Lemonade	-	finance	Workday	-	tech_software_saas
Levels	-	mental_health_wellness	Xero	-	finance_fintech
Liberty Mutual	-	finance	YouTube	-	media_streaming
Lime	-	automotive_transport	YouTube TV	-	media_streaming
Linear	-	tech_software_saas	Zapier	-	tech
LinkedIn	LinkedIn Jobs	jobs	Zendesk	-	tech_software_saas
Liquid IV	Liquid I.V.	health	ZipRecruiter	Zip Recruiter	jobs
LMNT	Element	health	ZocDoc	Zoc Doc	health
Loom	-	tech_software_saas	Zoom	-	tech_software_saas
Lululemon	-	ecommerce_retail_dtc	Zscaler	-	tech

Mishearing corrections (174 entries, from src/utils/constants.py SPONSOR_ALIASES). Applied post-transcription to normalize Whisper output toward the canonical sponsor name. Distinct from the aliases column above, which lists intentional alternative spellings (e.g. AG1 vs Athletic Greens); the entries below are mostly Whisper mishearings (e.g. a firm -> Affirm, xerox -> Xero). Laid out in three side-by-side groups, read top-to-bottom in each group.

Heard as	Normalized to	Heard as	Normalized to	Heard as	Normalized to
1 password	1Password	good-rx	GoodRx	patron	Patreon
8 sleep	Eight Sleep	green chef	Green Chef	pay tree on	Patreon
8-sleep	Eight Sleep	green-chef	Green Chef	perplexity ai	Perplexity
a firm	Affirm	greenchef	Green Chef	perplexity-ai	Perplexity
a g one	Athletic Greens	grub hub	Grubhub	policy genius	PolicyGenius
ag 1	Athletic Greens	grub-hub	Grubhub	policy-genius	PolicyGenius
ag one	Athletic Greens	harrys	Harry’s	pyura	Pura
ag1	Athletic Greens	head space	Headspace	ray con	Raycon
athlean x	Athlean-X	head-space	Headspace	ray-con	Raycon
athlean-x	Athlean-X	hello fresh	HelloFresh	re tool	Retool
athletic greens one	Athletic Greens	hello-fresh	HelloFresh	ro gain	Rogaine
athleticgreens	Athletic Greens	him's	Hims	ro-gaine	Rogaine
bet mgm	BetMGM	hims & hers	Hims & Hers	rocket money	Rocket Money
bet-mgm	BetMGM	hims and hers	Hims & Hers	rocket-money	Rocket Money
better help	BetterHelp	honey love	Honeylove	rocketlawyer	Rocket Lawyer
better-help	BetterHelp	honey-love	Honeylove	rocketmoney	Rocket Money
birch box	Birchbox	honeylove	Honeylove	rocketmortgage	Rocket Mortgage
birch-box	Birchbox	hub spot	HubSpot	seat geek	SeatGeek
bit warden	Bitwarden	hub-spot	HubSpot	seat-geek	SeatGeek
bit-warden	Bitwarden	hubs pot	HubSpot	shop a fly	Shopify
blueapron	Blue Apron	imperfect foods	Imperfect Foods	shop fly	Shopify
brecks	Brex	imperfectfoods	Imperfect Foods	shop ify	Shopify
butcher box	Butcher Box	insta cart	Instacart	simpli safe	SimpliSafe
butcher-box	Butcher Box	insta-cart	Instacart	simpli-safe	SimpliSafe
butcherbox	Butcher Box	l m n t	LMNT	simply safe	SimpliSafe
car max	CarMax	legal zoom	LegalZoom	sky scanner	Skyscanner
car-max	CarMax	legal-zoom	LegalZoom	sky-scanner	Skyscanner
cloud flare	Cloudflare	legalzoom	LegalZoom	so fi	SoFi
cloud-flare	Cloudflare	liquid i v	Liquid IV	so-fi	SoFi
co pilot	GitHub Copilot	liquid i.v.	Liquid IV	square space	Squarespace
co-pilot	GitHub Copilot	liquid iv	Liquid IV	square-space	Squarespace
copilot	GitHub Copilot	liquidiv	Liquid IV	stamp dot com	Stamps.com
creditkarma	Credit Karma	magic mind	Magic Mind	stitch fix	Stitch Fix
delete me	DeleteMe	magic spoon	Magic Spoon	stitch-fix	Stitch Fix
delete-me	DeleteMe	magicmind	Magic Mind	stitchfix	Stitch Fix
dollarshaveclub	Dollar Shave Club	magicspoon	Magic Spoon	stub hub	StubHub
door dash	DoorDash	master class	MasterClass	stub-hub	StubHub
door-dash	DoorDash	master-class	MasterClass	sub stack	Substack
draft kings	DraftKings	mercury bank	Mercury	sub-stack	Substack
draft-kings	DraftKings	mercury-bank	Mercury	thrive market	Thrive Market
eight-sleep	Eight Sleep	mint mobile	Mint Mobile	thrivemarket	Thrive Market
eightsleep	Eight Sleep	mint-mobile	Mint Mobile	transparent labs	Transparent Labs
element	LMNT	mintmobile	Mint Mobile	transparentlabs	Transparent Labs
every plate	EveryPlate	monarch money	Monarch Money	uber eats	Uber Eats
every-plate	EveryPlate	monarch-money	Monarch Money	uber-eats	Uber Eats
express vpn	ExpressVPN	monarchmoney	Monarch Money	ubereats	Uber Eats
express-vpn	ExpressVPN	my protein	Myprotein	ver cell	Vercel
fab fit fun	FabFitFun	my ro	Miro	ver sel	Vercel
fab-fit-fun	FabFitFun	myprotein	Myprotein	wealth front	Wealthfront
fan duel	FanDuel	net suite	NetSuite	wealth-front	Wealthfront
fan-duel	FanDuel	net-suite	NetSuite	woop	Whoop
game time	Gametime	nord vpn	NordVPN	xerox	Xero
game-time	Gametime	nord-vpn	NordVPN	zero	Xero
gametime	Gametime	one password	1Password	zip recruiter	ZipRecruiter
github-copilot	GitHub Copilot	one skin	OneSkin	zip-recruiter	ZipRecruiter
go puff	Gopuff	one-password	1Password	zoc doc	ZocDoc
go-puff	Gopuff	one-skin	OneSkin	zoc-doc	ZocDoc
good rx	GoodRx	p ninety x	P90X	zock doc	ZocDoc

Run Metadata

• Report generated: 2026-05-22T21:01:58Z
• Unique work units (current state, last-write-wins after retries): 37620
• Raw rows in calls.jsonl: 38251 (631 superseded by later retries; kept for audit)
• Successful: 37617
• Failed: 3
• Lifetime actual spend (sum of at-runtime costs, includes superseded rows): $299.3819
• Active pricing snapshot: 2026-05-20T19:59:56.229629Z