wfxr/code-minimap | ProgramBench

🛰 A high performance code minimap render.

313

Generated Behavioral Tests

99.7%

Best Score

Claude Opus 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)

25 runs

#	Model	Score	Cost	Calls
1	Claude Opus 5 (xhigh)	99.7%	$22.99	145	trace →
2	Muse Spark 1.3 (max)	99.4%	$1.59	157	trace →
3	Muse Spark 1.3 (xhigh)	99.4%	$0.66	140	trace →
4	Claude Opus 4.8 (xhigh)	99.4%	$8.64	77	trace →
5	GLM-5.2	99.0%	$11.42	147	trace →
6	Gemini 3.7 Flash	96.5%	$0.80	96	trace →
7	GPT 5.5 (xhigh)	95.8%	$4.65	48	trace →
8	Muse Spark 1.2 (xhigh)	94.2%	$1.50	211	trace →
9	Claude Opus 4.7 (xhigh)	91.7%	$9.90	139	trace →
10	Gemini 3.6 Flash	90.4%	$2.78	127	trace →
11	Claude Opus 4.7	88.8%	$9.23	184	trace →
12	GPT-5.6 Sol (xhigh)	86.6%	$3.21	26	trace →
13	Claude Opus 4.6	85.6%	$9.41	236	trace →
14	Gemini 3.5 Flash	85.3%	$1.83	118	trace →
15	GPT-5.6 Sol	84.7%	$0.80	13	trace →
16	GPT 5.5 (high)	82.7%	$3.01	36	trace →
17	Muse Spark 1.1 (xhigh)	79.9%	$1.27	246	trace →
18	GPT 5.5	75.7%	$0.88	13	trace →
19	Gemini 3.1 Pro	73.5%	$1.82	186	trace →
20	Claude Sonnet 4.6	71.6%	$14.48	348	trace →
21	Gemini 3 Flash	53.0%	$0.26	107	trace →
22	Claude Haiku 4.5	49.2%	$0.98	131	trace →
23	GPT 5.4	38.0%	$0.25	11	trace →
24	GPT 5 mini	19.2%	$0.09	43	trace →
25	GPT 5.4 mini	0.0%	$0.06	11	trace →

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