stacked-git/stgit | ProgramBench

Stacked Git

1,488

Generated Behavioral Tests

89.1%

Best Score

Claude Fable 5.1 (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 Fable 5.1 (xhigh)	89.1%	$122.75	468	trace →
2	Claude Opus 5.5 (xhigh)	88.3%	$85.35	635	trace →
3	Muse Wonder Full	85.7%	TBA	497	trace →
4	GPT 6 Astra (max)	83.9%	$15.04	63	trace →
5	Claude Opus 5 (xhigh)	82.8%	$108.71	470	trace →
6	Muse Wonder (xhigh)	75.3%	$33.45	353	trace →
7	Gemini 3.7 Flash	50.7%	$2.16	184	trace →
8	GPT 5.5 (high)	48.7%	$4.62	44	trace →
9	GPT 5.5 (xhigh)	46.4%	$7.28	80	trace →
10	Claude Opus 4.8 (xhigh)	46.4%	$30.01	214	trace →
11	GLM-5.2	40.9%	$20.78	219	trace →
12	Muse Spark 1.3 (xhigh)	40.6%	$1.10	197	trace →
13	Muse Spark 1.2 (xhigh)	40.0%	$0.77	107	trace →
14	GPT-5.6 Sol	39.0%	$0.99	17	trace →
15	GPT 5.5	37.7%	$1.13	18	trace →
16	Gemini 3.5 Flash	27.2%	$4.06	171	trace →
17	Claude Opus 4.7 (xhigh)	25.1%	$10.31	162	trace →
18	GPT 5 mini	20.0%	$0.02	14	trace →
19	GPT 5.4	16.3%	$0.23	8	trace →
20	Gemini 3 Flash	12.6%	$0.24	114	trace →
21	Claude Opus 4.7	10.1%	$0.56	25	trace →
22	Gemini 3.1 Pro	8.6%	$1.18	94	trace →
23	Claude Haiku 4.5	8.4%	$0.60	104	trace →
24	GPT 5.4 mini	7.1%	$0.02	9	trace →
25	Muse Spark 1.3 (max)	4.0%	$7.87	370	trace →
26	Gemini 3.6 Flash	4.0%	$2.04	91	trace →
27	Claude Sonnet 4.6	4.0%	$31.22	581	trace →
28	GPT-5.6 Sol (xhigh)	2.6%	$7.16	47	trace →
29	Muse Spark 1.1 (xhigh)	2.6%	$1.34	176	trace →
30	Claude Opus 4.6	0.0%	$1.93	213	trace →

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