ProofBench v1.1 · Open Analysis

#	Model	Effort	Score	Cost / task	Harness
1	Sonnet 5.5	max	100.0%	$0.558	–
2	Opus 5.5	max	100.0%	$0.962	–
3	Fable 5.1	max	100.0%	$2.70	–
4	AlephProver	–	100.0%	$9.35	–
5	GPT-6.1 Sol	max	99.0%	$0.233	–
6	GPT-6 Astra	max	99.0%	$1.71	–
7	Gemini 4	high	99.0%	$0.783	–
8	Opus 5	max	99.0%	$2.11	–
9	Fable 5	max	95.0%	$5.45	–
10	Wonder Full	max	94.0%	$0.856	–
11	Haiku 5.5	max	89.0%	$0.174	–
12	Kimi K3	max	87.0%	$1.65	–
13	Aristotle	–	86.0%	–	–
14	Wonder	max	84.0%	$1.31	–
15	GPT-6 Sol	max	83.0%	$0.476	–
16	GPT-5.6 Sol	max	83.0%	$1.44	–
17	Sonnet 5	max	77.0%	$1.37	–
18	Hy4	–	75.0%	$0.336	–
19	GPT-5.6 Terra	max	74.0%	$1.27	–
20	MiMo V2.6 Pro	high	70.0%	$0.245	–
21	GPT-6 Luna	max	64.0%	$0.042	–
22	MiMo V2.6 Flash	high	63.0%	$0.123	–
23	Ember-1	–	61.0%	$1.74	–
24	GPT-5.6 Luna	max	60.0%	$0.123	–
25	Gemini 3.7 Flash	high	58.0%	$0.559	–
26	Muse Spark 1.3	max	58.0%	$0.479	–
27	Qwen 3.8 Max	max	58.0%	$1.50	–
28	DeepSeek V4 Flash 0731	high	56.0%	$0.037	–
29	DeepSeek V4.1 Flash	high	54.0%	$0.126	–
30	Grok 4.6	high	51.0%	$0.759	–
31	DeepSeek V4 Pro 0813	max	50.0%	$0.062	–
32	GLM 5.3	max	49.0%	$2.08	–
33	Gemini 3.8 Flash	high	48.0%	$0.603	–
34	Muse Spark 1.2	xhigh	43.0%	$0.442	–
35	Step 5 Preview	–	42.0%	$0.715	–
36	Gemini 3.5 Flash	high	31.0%	$0.570	–
37	Grok 4.5	high	31.0%	$0.574	–
38	Grok 4.7	xhigh	26.0%	$0.788	–
39	Gemini 3.1 Pro	high	26.0%	$0.784	–
40	MiMo V2.5 Pro	xhigh	22.0%	$0.090	–
41	GLM 5.3 Flash	max	21.0%	$0.224	–
42	MiniMax M3	max	18.0%	$0.419	–
43	DeepSeek V4	max	16.0%	$0.034	–
44	MiMo V2.5	xhigh	16.0%	$0.057	–
45	Qwen 3.8 27B	xhigh	16.0%	$2.30	–
46	Mistral Large 4	high	10.0%	$0.642	–
47	Mistral Medium 3.5	high	9.0%	$1.30	–
48	Inkling Small	xhigh	6.0%	$0.075	–
49	Mercury 2.5	high	3.0%	$0.044	–
50	Inkling	max	0.0%	$0.201	–
51	Nemotron 3.5 Lightning	–	0.0%	$0.016	–