mkj/dropbear | ProgramBench

Dropbear SSH

667

Generated Behavioral Tests

83.2%

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)

30 runs

#	Model	Score	Cost	Calls
1	Muse Wonder Full	83.2%	TBA	624	trace →
2	GPT 6 Astra (max)	82.5%	$11.31	51	trace →
3	Claude Opus 5 (xhigh)	75.3%	$49.32	264	trace →
4	Claude Opus 4.8 (xhigh)	74.4%	$7.67	91	trace →
5	GLM-5.2	70.9%	$4.92	105	trace →
6	Gemini 3.7 Flash	70.8%	$2.25	127	trace →
7	Muse Wonder (xhigh)	68.7%	$47.93	480	trace →
8	GPT-5.6 Sol	67.8%	$0.65	14	trace →
9	GPT 5.5 (xhigh)	66.4%	$5.49	64	trace →
10	Gemini 3.6 Flash	66.1%	$11.00	239	trace →
11	Muse Spark 1.1 (xhigh)	65.1%	$0.15	100	trace →
12	GPT-5.6 Sol (xhigh)	62.7%	$4.29	44	trace →
13	Muse Spark 1.2 (xhigh)	60.7%	$0.78	102	trace →
14	GPT 5.4	58.9%	$0.15	10	trace →
15	Muse Spark 1.3 (max)	57.1%	$1.75	180	trace →
16	Muse Spark 1.3 (xhigh)	56.2%	$0.60	144	trace →
17	Claude Opus 4.7 (xhigh)	56.0%	$3.00	72	trace →
18	GPT 5.5 (high)	53.2%	$2.08	30	trace →
19	Claude Opus 4.6	52.5%	$8.15	152	trace →
20	Gemini 3.5 Flash	48.4%	$3.33	121	trace →
21	GPT 5.5	47.2%	$0.96	18	trace →
22	Gemini 3 Flash	45.7%	$0.19	56	trace →
23	Claude Opus 4.7	44.2%	$1.62	56	trace →
24	Claude Sonnet 4.6	42.7%	$67.81	600	trace →
25	Gemini 3.1 Pro	40.2%	$1.05	69	trace →
26	Claude Haiku 4.5	27.0%	$0.26	78	trace →
27	GPT 5.4 mini	22.3%	$0.02	7	trace →
28	GPT 5 mini	17.2%	$0.01	12	trace →
29	Claude Opus 5.5 (xhigh)	0.0%	$5.29	92	trace →
30	Claude Fable 5.1 (xhigh)	0.0%	$8.53	90	trace →

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