rust-ethereum/ethabi | ProgramBench

Encode and decode smart contract invocations

997

Generated Behavioral Tests

94.5%

Best Score

Claude Opus 4.8 (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 4.8 (xhigh)	94.5%	$24.18	175	trace →
2	Muse Spark 1.3 (xhigh)	94.5%	$1.83	271	trace →
3	Muse Spark 1.3 (max)	93.2%	$2.84	227	trace →
4	Muse Spark 1.1 (xhigh)	91.9%	$0.49	308	trace →
5	Claude Opus 4.7 (xhigh)	90.9%	$14.23	225	trace →
6	GPT-5.6 Sol (xhigh)	89.9%	$2.05	18	trace →
7	GLM-5.2	89.9%	$14.89	163	trace →
8	Claude Opus 4.7	88.7%	$13.93	246	trace →
9	Gemini 3.5 Flash	87.9%	$7.09	234	trace →
10	Claude Sonnet 4.6	87.6%	$12.55	407	trace →
11	Muse Spark 1.2 (xhigh)	86.5%	$1.26	142	trace →
12	GPT 5.5 (xhigh)	86.3%	$6.65	73	trace →
13	Gemini 3.6 Flash	85.8%	$4.08	147	trace →
14	GPT 5.5 (high)	83.9%	$4.62	44	trace →
15	Claude Opus 4.6	81.6%	$10.26	257	trace →
16	GPT-5.6 Sol	77.7%	$0.71	12	trace →
17	GPT 5.5	72.2%	$0.98	15	trace →
18	Claude Haiku 4.5	50.9%	$1.25	159	trace →
19	GPT 5.4	48.2%	$0.32	12	trace →
20	GPT 5 mini	29.7%	$0.03	25	trace →
21	Gemini 3.7 Flash	11.5%	$2.12	132	trace →
22	Gemini 3.1 Pro	7.8%	$2.23	160	trace →
23	Gemini 3 Flash	7.8%	$0.23	89	trace →
24	Claude Opus 5 (xhigh)	0.0%	$3.07	46	trace →
25	GPT 5.4 mini	0.0%	$0.03	14	trace →

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