yoav-lavi/melody | ProgramBench

Melody is a language that compiles to regular expressions and aims to be more readable and maintainable

1,205

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%	$19.30	255	trace →
2	Muse Wonder (xhigh)	99.3%	$40.09	375	trace →
3	Muse Wonder Full	99.0%	TBA	279	trace →
4	GPT 6 Astra (max)	94.9%	$10.23	51	trace →
5	Claude Fable 5.1 (xhigh)	93.2%	$27.34	212	trace →
6	Claude Opus 5 (xhigh)	93.2%	$22.60	175	trace →
7	Muse Spark 1.3 (xhigh)	92.9%	$2.91	312	trace →
8	Claude Opus 4.8 (xhigh)	90.5%	$31.56	154	trace →
9	Muse Spark 1.3 (max)	90.0%	$3.69	236	trace →
10	Gemini 3.7 Flash	88.9%	$1.30	136	trace →
11	Claude Opus 4.7 (xhigh)	87.4%	$17.22	215	trace →
12	Muse Spark 1.2 (xhigh)	85.8%	$2.64	283	trace →
13	GPT 5.5 (xhigh)	84.9%	$4.22	56	trace →
14	Claude Opus 4.7	78.9%	$5.37	135	trace →
15	GPT-5.6 Sol (xhigh)	78.3%	$2.19	21	trace →
16	GLM-5.2	78.2%	$22.03	278	trace →
17	GPT-5.6 Sol	77.4%	$1.03	16	trace →
18	Claude Opus 4.6	74.4%	$12.05	325	trace →
19	GPT 5.5 (high)	71.5%	$3.79	54	trace →
20	Gemini 3.6 Flash	66.4%	$3.80	180	trace →
21	GPT 5.5	62.3%	$1.25	21	trace →
22	Gemini 3 Flash	62.3%	$0.28	138	trace →
23	GPT 5.4	34.2%	$0.45	19	trace →
24	Claude Haiku 4.5	31.4%	$0.98	131	trace →
25	GPT 5.4 mini	6.4%	$0.03	9	trace →
26	Gemini 3.1 Pro	6.1%	$2.10	203	trace →
27	Claude Sonnet 4.6	3.4%	$16.67	502	trace →
28	GPT 5 mini	3.4%	$0.04	25	trace →
29	Muse Spark 1.1 (xhigh)	1.0%	$0.60	191	trace →
30	Gemini 3.5 Flash	0.0%	$5.11	280	trace →

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