FFmpeg/FFmpeg | ProgramBench

Mirror of https://git.ffmpeg.org/ffmpeg.git

3,041

Generated Behavioral Tests

13.9%

Best Score

GPT 6 Astra (max)

Results by Model

Hover for details · Click a model to open it · Drag to zoom the x-axis · Shift-drag to pan · The line marks the Pareto frontier (best score per cost)

30 runs

#	Model	Score	Cost	Calls
1	GPT 6 Astra (max)	13.9%	$14.70	74	trace →
2	Muse Wonder Full	13.0%	TBA	512	trace →
3	Muse Wonder (xhigh)	12.6%	$24.82	309	trace →
4	Claude Fable 5.1 (xhigh)	10.5%	$36.68	178	trace →
5	Muse Spark 1.1 (xhigh)	10.4%	$0.29	230	trace →
6	Muse Spark 1.2 (xhigh)	8.3%	$0.25	100	trace →
7	Gemini 3.5 Flash	8.1%	$1.21	77	trace →
8	GPT 5.5 (xhigh)	7.0%	$3.59	32	trace →
9	GPT-5.6 Sol (xhigh)	6.9%	$4.03	25	trace →
10	GLM-5.2	6.8%	$6.12	92	trace →
11	GPT 5.5	6.1%	$1.16	15	trace →
12	Gemini 3.1 Pro	5.3%	$0.92	52	trace →
13	Muse Spark 1.3 (xhigh)	4.8%	$2.94	320	trace →
14	GPT 5.4	4.6%	$0.39	12	trace →
15	Claude Sonnet 4.6	4.5%	$12.51	273	trace →
16	GPT 5.5 (high)	4.2%	$2.60	28	trace →
17	GPT 5.4 mini	3.9%	$0.03	13	trace →
18	Claude Opus 4.8 (xhigh)	3.7%	$12.79	76	trace →
19	Gemini 3 Flash	3.5%	$0.38	89	trace →
20	Claude Opus 4.7	3.4%	$1.07	35	trace →
21	Claude Haiku 4.5	2.8%	$0.48	103	trace →
22	Claude Opus 4.7 (xhigh)	2.4%	$1.50	32	trace →
23	Muse Spark 1.3 (max)	2.2%	$17.93	524	trace →
24	GPT 5 mini	2.2%	$0.02	15	trace →
25	GPT-5.6 Sol	2.2%	$0.45	8	trace →
26	Gemini 3.6 Flash	2.1%	$4.60	121	trace →
27	Gemini 3.7 Flash	2.1%	$0.60	66	trace →
28	Claude Opus 5.5 (xhigh)	0.0%	$66.17	392	trace →
29	Claude Opus 5 (xhigh)	0.0%	$6.33	53	trace →
30	Claude Opus 4.6	0.0%	$2.06	221	trace →

Click a row to replay how that model rebuilt this program, or the model name to open its full run