chirlu/sox | ProgramBench

SoX, Swiss Army knife of sound processing

1,196

Generated Behavioral Tests

87.8%

Best Score

Claude Opus 5.5 (xhigh)

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	Claude Opus 5.5 (xhigh)	87.8%	$51.29	286	trace →
2	Claude Fable 5.1 (xhigh)	80.3%	$117.06	330	trace →
3	GPT 6 Astra (max)	74.2%	$15.45	74	trace →
4	Muse Wonder (xhigh)	68.6%	$44.10	417	trace →
5	Muse Wonder Full	64.6%	TBA	644	trace →
6	Muse Spark 1.3 (xhigh)	51.7%	$1.61	239	trace →
7	GPT-5.6 Sol (xhigh)	51.6%	$6.88	46	trace →
8	Claude Opus 5 (xhigh)	49.2%	$60.63	253	trace →
9	Gemini 3.7 Flash	47.7%	$1.11	100	trace →
10	GPT 5.5 (high)	38.9%	$4.14	39	trace →
11	Claude Opus 4.6	38.0%	$10.13	233	trace →
12	GPT-5.6 Sol	36.9%	$1.20	15	trace →
13	GLM-5.2	32.1%	$20.09	193	trace →
14	GPT 5.5	29.5%	$1.30	16	trace →
15	Claude Sonnet 4.6	29.3%	$13.13	287	trace →
16	Claude Opus 4.8 (xhigh)	24.9%	$21.77	128	trace →
17	GPT 5.4	13.4%	$0.23	7	trace →
18	Gemini 3 Flash	8.6%	$0.29	66	trace →
19	Claude Opus 4.7 (xhigh)	7.8%	$7.08	125	trace →
20	Gemini 3.1 Pro	7.0%	$1.15	53	trace →
21	Claude Opus 4.7	3.7%	$1.25	49	trace →
22	GPT 5 mini	3.6%	$0.03	10	trace →
23	Claude Haiku 4.5	3.1%	$0.86	125	trace →
24	GPT 5.4 mini	1.3%	$0.04	16	trace →
25	Muse Spark 1.2 (xhigh)	1.3%	$0.24	87	trace →
26	Gemini 3.5 Flash	1.3%	$4.31	132	trace →
27	Muse Spark 1.3 (max)	1.1%	$6.34	312	trace →
28	Gemini 3.6 Flash	1.1%	$2.72	106	trace →
29	Muse Spark 1.1 (xhigh)	1.1%	$0.27	120	trace →
30	GPT 5.5 (xhigh)	n/a	$79.42	1,000	trace →

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