madler/pigz | ProgramBench

A parallel implementation of gzip for modern multi-processor, multi-core machines.

831

Generated Behavioral Tests

99.8%

Best Score

Muse Wonder Full

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)

28 runs

#	Model	Score	Cost	Calls
1	Muse Wonder Full	99.8%	TBA	436	trace →
2	GPT 6 Astra (max)	99.6%	$11.00	50	trace →
3	Claude Opus 5 (xhigh)	98.6%	$24.31	181	trace →
4	Muse Wonder (xhigh)	98.4%	$36.20	375	trace →
5	Muse Spark 1.3 (max)	98.1%	$4.65	327	trace →
6	Muse Spark 1.3 (xhigh)	97.1%	$1.85	258	trace →
7	GPT 5.5 (xhigh)	94.9%	$10.57	91	trace →
8	GPT-5.6 Sol (xhigh)	93.6%	$4.26	36	trace →
9	Claude Opus 4.7 (xhigh)	92.2%	$8.06	152	trace →
10	Muse Spark 1.2 (xhigh)	91.8%	$1.02	129	trace →
11	GPT-5.6 Sol	90.6%	$1.02	22	trace →
12	GPT 5.5 (high)	89.4%	$2.86	38	trace →
13	Gemini 3.5 Flash	85.1%	$3.72	161	trace →
14	GPT 5.5	84.2%	$1.03	17	trace →
15	Claude Opus 4.7	83.2%	$4.53	149	trace →
16	GLM-5.2	80.6%	$39.11	213	trace →
17	GPT 5.4	79.1%	$0.23	10	trace →
18	Claude Sonnet 4.6	74.8%	$11.64	343	trace →
19	Claude Haiku 4.5	73.4%	$0.47	84	trace →
20	Gemini 3.1 Pro	66.4%	$1.10	88	trace →
21	Gemini 3.7 Flash	50.1%	$1.81	112	trace →
22	Claude Opus 4.8 (xhigh)	47.9%	$21.11	140	trace →
23	Gemini 3 Flash	29.6%	$0.14	37	trace →
24	GPT 5 mini	10.6%	$0.01	8	trace →
25	Muse Spark 1.1 (xhigh)	2.3%	$0.16	92	trace →
26	Claude Opus 5.5 (xhigh)	0.0%	$8.46	102	trace →
27	Claude Opus 4.6	0.0%	$1.63	215	trace →
28	GPT 5.4 mini	0.0%	$0.02	13	trace →

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