cslarsen/jp2a | ProgramBench

Converts jpg images to ASCII

631

Generated Behavioral Tests

100.0%

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)	100.0%	$30.40	253	trace →
2	Claude Fable 5.1 (xhigh)	99.4%	$47.05	206	trace →
3	GPT 6 Astra (max)	98.4%	$10.97	44	trace →
4	Muse Wonder Full	95.9%	TBA	260	trace →
5	Claude Opus 5 (xhigh)	93.2%	$105.34	269	trace →
6	Muse Wonder (xhigh)	92.7%	$29.03	303	trace →
7	GPT-5.6 Sol (xhigh)	88.1%	$7.48	54	trace →
8	Claude Opus 4.8 (xhigh)	75.0%	$31.35	156	trace →
9	Muse Spark 1.3 (xhigh)	62.3%	$2.68	299	trace →
10	GPT 5.5 (xhigh)	61.3%	$8.75	97	trace →
11	Gemini 3.7 Flash	58.2%	$2.89	178	trace →
12	Gemini 3.5 Flash	57.8%	$8.01	227	trace →
13	Muse Spark 1.1 (xhigh)	57.2%	$0.26	221	trace →
14	Claude Opus 4.7	56.1%	$16.11	239	trace →
15	Gemini 3.6 Flash	55.9%	$4.24	167	trace →
16	GPT-5.6 Sol	53.6%	$1.99	26	trace →
17	GPT 5.5	45.0%	$1.50	28	trace →
18	GPT 5.5 (high)	44.4%	$5.11	46	trace →
19	Muse Spark 1.2 (xhigh)	44.4%	$0.43	104	trace →
20	Claude Sonnet 4.6	42.2%	$45.93	495	trace →
21	Claude Opus 4.6	42.0%	$13.07	192	trace →
22	Gemini 3 Flash	33.3%	$0.31	68	trace →
23	Claude Haiku 4.5	30.0%	$1.24	162	trace →
24	Gemini 3.1 Pro	26.0%	$2.21	100	trace →
25	GPT 5.4	25.2%	$0.43	15	trace →
26	Claude Opus 4.7 (xhigh)	19.2%	$15.72	169	trace →
27	GLM-5.2	14.1%	$24.47	176	trace →
28	GPT 5.4 mini	14.1%	$0.05	37	trace →
29	GPT 5 mini	4.6%	$0.01	12	trace →
30	Muse Spark 1.3 (max)	0.0%	$15.57	457	trace →

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