ArthurSonzogni/json-tui

A JSON terminal UI made in C++

755

Generated Behavioral Tests

93.5%

Best Score

GPT-5.6 Sol (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	GPT-5.6 Sol (xhigh)	93.5%	$4.52	27	trace →
2	Claude Opus 5 (xhigh)	92.2%	$74.35	208	trace →
3	GPT 5.5 (xhigh)	90.2%	$5.22	43	trace →
4	Muse Spark 1.1 (xhigh)	90.1%	$0.24	174	trace →
5	GPT 5.5 (high)	89.5%	$3.69	34	trace →
6	Muse Spark 1.2 (xhigh)	87.5%	$0.11	172	trace →
7	Gemini 3.5 Flash	84.5%	$4.27	116	trace →
8	Claude Opus 4.8 (xhigh)	83.0%	$17.18	102	trace →
9	GPT-5.6 Sol	74.3%	$1.26	17	trace →
10	GLM-5.2	73.6%	$20.94	200	trace →
11	Claude Opus 4.6	71.0%	$13.34	298	trace →
12	Gemini 3.6 Flash	69.3%	$4.77	117	trace →
13	GPT 5.5	68.6%	$0.96	14	trace →
14	Muse Spark 1.3 (max)	64.8%	$13.92	578	trace →
15	Gemini 3.7 Flash	63.8%	$1.23	71	trace →
16	Muse Spark 1.3 (xhigh)	55.0%	$1.49	226	trace →
17	Claude Haiku 4.5	52.6%	$0.52	81	trace →
18	Claude Sonnet 4.6	52.3%	$27.61	444	trace →
19	Gemini 3.1 Pro	46.8%	$1.11	60	trace →
20	Claude Opus 4.7	45.0%	$1.07	43	trace →
21	GPT 5.4 mini	44.4%	$0.03	7	trace →
22	GPT 5.4	39.6%	$0.10	7	trace →
23	GPT 5 mini	35.6%	$0.01	10	trace →
24	Gemini 3 Flash	19.5%	$0.20	59	trace →
25	Claude Opus 4.7 (xhigh)	9.4%	$4.62	116	trace →

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