lua/lua | ProgramBench

← Back to leaderboard · Show all task instances

A copy of the Lua development repository, as seen by the Lua team. Mirrored irregularly. All communication should be through the Lua mailing list https://www.lua.org/lua-l.html

9,908 c medium

1,338

Generated Behavioral Tests

98.7%

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)

Per-Model Results

30 runs

#	Model	Score	Cost	Calls
1	Claude Opus 5.5 (xhigh)	98.7%	$63.77	385	trace →
2	Claude Fable 5.1 (xhigh)	98.1%	$57.18	187	trace →
3	GPT 6 Astra (max)	97.7%	$11.53	51	trace →
4	Muse Wonder (xhigh)	95.2%	$32.01	270	trace →
5	Claude Opus 5 (xhigh)	94.7%	$50.75	204	trace →
6	Muse Wonder Full	93.5%	TBA	226	trace →
7	Muse Spark 1.3 (xhigh)	85.7%	$1.47	192	trace →
8	GPT-5.6 Sol (xhigh)	85.1%	$5.63	42	trace →
9	Muse Spark 1.3 (max)	81.8%	$5.54	282	trace →
10	Claude Opus 4.8 (xhigh)	81.5%	$23.23	148	trace →
11	Gemini 3.7 Flash	78.9%	$1.18	89	trace →
12	GPT-5.6 Sol	76.5%	$1.63	21	trace →
13	GPT 5.5 (xhigh)	76.2%	$5.54	63	trace →
14	GPT 5.5 (high)	66.1%	$4.34	50	trace →
15	Muse Spark 1.2 (xhigh)	44.8%	$2.98	229	trace →
16	Claude Opus 4.7	43.1%	$4.41	95	trace →
17	GPT 5.5	43.0%	$1.02	14	trace →
18	GLM-5.2	41.6%	$78.66	384	trace →
19	Claude Sonnet 4.6	34.2%	$138.99	993	trace →
20	GPT 5.4	17.8%	$0.20	13	trace →
21	Gemini 3.1 Pro	13.9%	$0.80	64	trace →
22	GPT 5 mini	5.1%	$0.02	16	trace →
23	GPT 5.4 mini	4.5%	$0.20	265	trace →
24	Gemini 3 Flash	3.4%	$0.17	112	trace →
25	Gemini 3.6 Flash	2.2%	$2.84	147	trace →
26	Claude Opus 4.7 (xhigh)	2.2%	$18.58	197	trace →
27	Muse Spark 1.1 (xhigh)	2.2%	$0.95	200	trace →
28	Claude Haiku 4.5	2.2%	$1.02	157	trace →
29	Gemini 3.5 Flash	1.6%	$4.85	147	trace →
30	Claude Opus 4.6	0.0%	$9.84	315	trace →

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