sirwart/ripsecrets | ProgramBench

A command-line tool to prevent committing secret keys into your source code

611

Generated Behavioral Tests

97.4%

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)	97.4%	$24.07	249	trace →
2	Muse Wonder (xhigh)	96.6%	$34.83	325	trace →
3	GPT 6 Astra (max)	95.7%	$9.03	46	trace →
4	Muse Wonder Full	94.8%	TBA	339	trace →
5	Muse Spark 1.3 (max)	94.6%	$4.63	251	trace →
6	Claude Fable 5.1 (xhigh)	94.3%	$33.19	171	trace →
7	Muse Spark 1.3 (xhigh)	94.1%	$1.87	262	trace →
8	GPT-5.6 Sol (xhigh)	93.5%	$4.67	42	trace →
9	GPT 5.5 (high)	91.2%	$4.05	52	trace →
10	GPT-5.6 Sol	89.7%	$0.60	12	trace →
11	GPT 5.5	88.7%	$0.74	16	trace →
12	GPT 5.5 (xhigh)	87.7%	$6.58	77	trace →
13	Gemini 3.6 Flash	85.1%	$4.26	139	trace →
14	Gemini 3.7 Flash	81.3%	$1.35	97	trace →
15	Claude Opus 5 (xhigh)	79.7%	$63.35	234	trace →
16	Muse Spark 1.1 (xhigh)	78.2%	$0.62	101	trace →
17	Claude Opus 4.7 (xhigh)	77.4%	$15.13	210	trace →
18	GLM-5.2	74.3%	$29.05	281	trace →
19	Muse Spark 1.2 (xhigh)	74.1%	$1.07	150	trace →
20	Claude Opus 4.7	72.8%	$7.44	156	trace →
21	Claude Opus 4.8 (xhigh)	66.6%	$22.96	141	trace →
22	Gemini 3.5 Flash	64.6%	$6.24	147	trace →
23	Claude Opus 4.6	63.8%	$16.13	268	trace →
24	Gemini 3 Flash	59.9%	$0.39	129	trace →
25	GPT 5.4	46.3%	$0.24	12	trace →
26	GPT 5 mini	42.9%	$0.01	16	trace →
27	Claude Haiku 4.5	18.0%	$1.51	186	trace →
28	Claude Sonnet 4.6	7.4%	$65.81	845	trace →
29	Gemini 3.1 Pro	0.0%	$2.14	115	trace →
30	GPT 5.4 mini	0.0%	$0.04	10	trace →

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