jhspetersson/fselect | ProgramBench

Find files with SQL-like queries

3,115

Generated Behavioral Tests

95.0%

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)	95.0%	$57.85	435	trace →
2	Muse Wonder (xhigh)	93.1%	$44.53	459	trace →
3	GPT 6 Astra (max)	92.1%	$13.28	62	trace →
4	Claude Fable 5.1 (xhigh)	90.5%	$91.50	379	trace →
5	Claude Opus 5 (xhigh)	87.0%	$98.31	469	trace →
6	GPT 5.5 (xhigh)	76.4%	$11.81	82	trace →
7	Muse Spark 1.3 (xhigh)	75.3%	$1.20	188	trace →
8	GPT-5.6 Sol (xhigh)	75.1%	$6.17	45	trace →
9	Claude Opus 4.8 (xhigh)	72.6%	$22.17	208	trace →
10	GLM-5.2	71.0%	$23.96	252	trace →
11	GPT 5.5 (high)	64.9%	$3.56	39	trace →
12	GPT 5.5	57.3%	$1.57	18	trace →
13	Muse Spark 1.2 (xhigh)	54.3%	$2.70	254	trace →
14	Claude Opus 4.6	44.0%	$9.78	262	trace →
15	Gemini 3.5 Flash	42.4%	$9.72	275	trace →
16	GPT-5.6 Sol	39.9%	$1.68	19	trace →
17	Claude Opus 4.7	29.3%	$2.52	74	trace →
18	Gemini 3.1 Pro	24.1%	$0.89	62	trace →
19	Muse Wonder Full	23.2%	TBA	351	trace →
20	GPT 5.4 mini	20.3%	$0.05	9	trace →
21	GPT 5.4	20.3%	$0.43	12	trace →
22	Muse Spark 1.3 (max)	16.2%	$5.51	354	trace →
23	Claude Haiku 4.5	9.9%	$0.51	113	trace →
24	Gemini 3.6 Flash	6.2%	$5.44	161	trace →
25	Claude Sonnet 4.6	2.1%	$20.11	479	trace →
26	Gemini 3.7 Flash	1.7%	$2.16	162	trace →
27	Muse Spark 1.1 (xhigh)	1.7%	$1.23	190	trace →
28	Claude Opus 4.7 (xhigh)	1.4%	$7.60	126	trace →
29	Gemini 3 Flash	0.9%	$0.24	89	trace →
30	GPT 5 mini	0.0%	$0.07	28	trace →

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