bellard/quickjs | ProgramBench

Public repository of the QuickJS Javascript Engine.

3,034

Generated Behavioral Tests

91.8%

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)	91.8%	$77.69	425	trace →
2	Claude Fable 5.1 (xhigh)	91.1%	$106.74	358	trace →
3	Muse Wonder Full	87.4%	TBA	493	trace →
4	GPT 6 Astra (max)	79.6%	$16.50	55	trace →
5	Muse Wonder (xhigh)	76.9%	$32.05	333	trace →
6	Muse Spark 1.3 (max)	47.3%	$5.47	300	trace →
7	Claude Opus 4.8 (xhigh)	46.8%	$18.15	110	trace →
8	GPT-5.6 Sol (xhigh)	35.8%	$6.17	43	trace →
9	Muse Spark 1.3 (xhigh)	32.8%	$2.43	295	trace →
10	Claude Opus 4.7 (xhigh)	31.0%	$31.65	261	trace →
11	GLM-5.2	28.6%	$49.19	351	trace →
12	GPT 5.5 (xhigh)	21.3%	$7.83	76	trace →
13	Gemini 3.7 Flash	20.5%	$1.25	96	trace →
14	GPT 5.5 (high)	10.9%	$3.51	30	trace →
15	GPT-5.6 Sol	7.9%	$1.22	14	trace →
16	Gemini 3.6 Flash	7.2%	$3.67	116	trace →
17	GPT 5.5	4.0%	$1.67	26	trace →
18	Muse Spark 1.1 (xhigh)	3.6%	$0.29	331	trace →
19	Claude Opus 4.6	3.6%	$16.39	248	trace →
20	GPT 5.4	1.2%	$0.21	13	trace →
21	Claude Opus 4.7	1.2%	$0.52	18	trace →
22	Gemini 3.1 Pro	1.0%	$1.30	119	trace →
23	Claude Sonnet 4.6	0.0%	$82.98	995	trace →
24	Claude Haiku 4.5	0.0%	$0.55	124	trace →
25	Claude Opus 5 (xhigh)	0.0%	$1.59	25	trace →
26	Muse Spark 1.2 (xhigh)	0.0%	$0.12	135	trace →
27	Gemini 3.5 Flash	0.0%	$3.24	160	trace →
28	Gemini 3 Flash	0.0%	$0.44	96	trace →
29	GPT 5.4 mini	0.0%	$0.01	8	trace →
30	GPT 5 mini	0.0%	$0.01	10	trace →

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