yassinebridi/serpl | ProgramBench

A simple terminal UI for search and replace, ala VS Code.

417

Generated Behavioral Tests

98.8%

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)	98.8%	$32.97	326	trace →
2	Claude Fable 5.1 (xhigh)	97.6%	$59.36	364	trace →
3	Claude Opus 5 (xhigh)	95.0%	$84.56	440	trace →
4	Muse Wonder Full	90.2%	TBA	359	trace →
5	Muse Wonder (xhigh)	89.7%	$75.73	603	trace →
6	GPT 6 Astra (max)	85.6%	$8.26	54	trace →
7	GPT-5.6 Sol (xhigh)	72.9%	$7.80	50	trace →
8	Muse Spark 1.3 (max)	71.0%	$10.01	427	trace →
9	Claude Opus 4.8 (xhigh)	69.5%	$20.65	180	trace →
10	Muse Spark 1.3 (xhigh)	69.3%	$1.54	265	trace →
11	Claude Opus 4.7 (xhigh)	69.1%	$2.51	79	trace →
12	Muse Spark 1.1 (xhigh)	66.7%	$0.49	146	trace →
13	GPT 5.5 (xhigh)	65.9%	$6.46	83	trace →
14	GPT 5.5 (high)	65.7%	$4.18	61	trace →
15	GLM-5.2	65.7%	$19.82	261	trace →
16	Gemini 3.7 Flash	65.0%	$4.26	167	trace →
17	Claude Opus 4.7	64.5%	$2.92	96	trace →
18	GPT 5.5	64.5%	$0.91	22	trace →
19	GPT-5.6 Sol	63.3%	$0.79	12	trace →
20	GPT 5.4	60.9%	$0.36	13	trace →
21	Claude Haiku 4.5	57.6%	$0.64	95	trace →
22	Muse Spark 1.2 (xhigh)	56.8%	$0.98	148	trace →
23	Gemini 3.5 Flash	56.4%	$7.57	168	trace →
24	Gemini 3.1 Pro	56.1%	$5.31	134	trace →
25	Gemini 3 Flash	48.0%	$0.44	116	trace →
26	GPT 5 mini	44.1%	$0.01	10	trace →
27	GPT 5.4 mini	43.9%	$0.03	15	trace →
28	Claude Opus 4.6	43.6%	$12.94	210	trace →
29	Claude Sonnet 4.6	7.0%	$17.28	310	trace →
30	Gemini 3.6 Flash	6.7%	$3.44	139	trace →

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