jqlang/jq | ProgramBench

Command-line JSON processor

6,072

Generated Behavioral Tests

98.4%

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)	98.4%	$78.04	259	trace →
2	Claude Opus 5.5 (xhigh)	97.3%	$55.86	319	trace →
3	Claude Opus 5 (xhigh)	94.7%	$100.84	341	trace →
4	GPT 6 Astra (max)	93.8%	$7.32	37	trace →
5	Muse Wonder Full	93.7%	TBA	403	trace →
6	Muse Spark 1.1 (xhigh)	91.3%	$0.35	119	trace →
7	GPT 5.4	90.1%	$0.15	140	trace →
8	Muse Wonder (xhigh)	89.5%	$51.50	508	trace →
9	Muse Spark 1.3 (max)	83.2%	$5.87	399	trace →
10	GPT-5.6 Sol (xhigh)	81.3%	$6.05	44	trace →
11	GPT 5.5 (xhigh)	75.4%	$15.55	148	trace →
12	Muse Spark 1.3 (xhigh)	72.1%	$2.89	327	trace →
13	Claude Opus 4.7	63.9%	$2.69	80	trace →
14	GPT-5.6 Sol	62.2%	$1.27	17	trace →
15	GPT 5.5 (high)	60.0%	$2.38	26	trace →
16	Gemini 3.6 Flash	56.1%	$7.69	170	trace →
17	Muse Spark 1.2 (xhigh)	39.5%	$0.82	172	trace →
18	GPT 5.5	31.5%	$0.67	13	trace →
19	Gemini 3.5 Flash	30.0%	$4.38	192	trace →
20	Claude Haiku 4.5	21.6%	$0.92	111	trace →
21	Gemini 3 Flash	18.7%	$0.21	46	trace →
22	Gemini 3.1 Pro	16.1%	$0.72	60	trace →
23	GPT 5 mini	7.8%	$0.01	10	trace →
24	GLM-5.2	1.1%	$59.61	268	trace →
25	Gemini 3.7 Flash	1.1%	$2.36	192	trace →
26	Claude Opus 4.7 (xhigh)	1.1%	$12.97	153	trace →
27	Claude Opus 4.6	1.1%	$18.26	384	trace →
28	Claude Sonnet 4.6	1.0%	$4.40	289	trace →
29	Claude Opus 4.8 (xhigh)	1.0%	$40.74	179	trace →
30	GPT 5.4 mini	1.0%	$0.02	15	trace →

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