Benchmark Almanac

ProofBench v1.1

Claude Fable 5.1 100% leads since September 2026
Claude Fable 5.1 100% leads since September 2026
Only in system cards
Version 1.1

The official leaderboard hasn’t scored a model released after Claude Fable 5.2 (November 2026). Labs report their own numbers for one newer model: Inkling 2 Mini at 18%.

Models write Lean 4 proofs of graduate-level theorems from paired natural-language and formal statements. Scored by the share the Lean checker accepts.

Vals says scores are not comparable with earlier revisions. Four systems tied at 100% on October 29th, 2026 among 54, so it is saturated.

Made by
Vals AI
Released
August 2026
Kind
Formal proofs
Level
Expert
Scored
Checked automatically
Answers
Written proof
Tasks
Public, plus a private test set
Measures
Capability
Language
English
In and out
Text, Code → Code
Interaction
Single turn
Run by
A host
Based on
ProofBench
Strongest model tested
Claude Fable 5.2 #2 on the capability index
Last in a system card
November 2026
Last model added
November 2026
Scores refreshed
November 19th, 2026
54
Entries on its own leaderboard, 52 models we track
1
System card reports it, from 1 lab

4 models share the top score of 100% 4 models share the top score of 100%

Best score per entry from the benchmark's own leaderboard, read on November 19th, 2026. Where a model ran with several settings or harnesses, we keep its best. Every run is in the next chart. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. This view shows only models we track, each at its best.

Claude Fable 5.1 leads with 100%, claims included

Best score per model from every source, including labs’ own numbers and claims Epoch AI collected from reports and papers. Labs pick their own settings, so a claim can beat an independent run of the same model. Striped bars are claims, and dotted bars are reruns by Epoch AI or another board.

  1. 1 AlephProver 100%
  2. 2 Claude Fable 5.1 100%
  3. 3 Claude Opus 5.5 100%
  4. 4 Claude Sonnet 5.5 100%
  5. 5 Claude Fable 5.2 100%
  6. 6 Claude Opus 5 99%
  7. 7 GPT-6 Astra 99%
  8. 8 GPT-6.1 Sol 99%
  9. 9 Gemini 4 Argon 99%
  10. 10 Claude Fable 5 95%
  11. 11 Muse Wonder Full 94%
  12. 12 Kimi K3.1 Pro 91%
  13. 13 Claude Haiku 5.5 89%
  14. 14 Kimi K3 87%
  15. 15 Aristotle 86%
  16. 16 Muse Wonder 84%
  17. 17 GPT-5.6 Sol 83%
  18. 18 GPT-6 Sol 83%
  19. 19 Claude Sonnet 5 77%
  20. 20 Tencent Hy4 preview 75%
  21. 21 GPT-5.6 Terra 74%
  22. 22 MiMo-V2.6 Pro 70%
  23. 23 GPT-6 Luna 64%
  24. 24 MiMo-V2.6 Flash 63%
  25. 25 Ember-1 61%
  26. 26 GPT-5.6 Luna 60%
  27. 27 Qwen 3.8 Max 58%
  28. 28 Gemini 3.7 Flash 58%
  29. 29 Muse Spark 1.3 max 58%
  30. 30 DeepSeek V4 Flash 0731 56%
  31. 31 DeepSeek V4.1 Flash 54%
  32. 32 Grok 4.6 51%
  33. 33 DeepSeek V4 Pro 0813 50%
  34. 34 GLM-5.3 49%
  35. 35 Gemini 3.8 Flash 48%
  36. 36 Muse Spark 1.2 43%
  37. 37 Step 5 Preview 42%
  38. 38 Gemini 3.5 Flash 31%
  39. 39 Grok 4.5 31%
  40. 40 Gemini 3.1 Pro 26%
  41. 41 Grok 4.7 26%
  42. 42 MiMo-V2.5-Pro 22%
  43. 43 GLM-5.3-Flash 21%
  44. 44 MiniMax-M3 18%
  45. 45 Beam 1.0 18%
  46. 46 MiMo-V2.5 16%
  47. 47 DeepSeek-V4-Pro 16%
  48. 48 Qwen 3.8 27B 16%
  49. 49 Mistral Large 4 Preview 10%
  50. 50 Mistral Medium 3.5 9.00%
  51. 51 Inkling-Small 6.00%
  52. 52 Mercury 2.5 3.00%
  53. 53 Inkling 0%
  54. 54 Nemotron 3.5 Lightning 30B-A3B 0%

Source: Official ProofBench v1.1 leaderboard

Source: Official ProofBench v1.1 leaderboard

Sources: Official ProofBench v1.1 leaderboard and labs’ model cards

#EntryScore ± 95% CISettingDateSource
1 AlephProver 100% Official leaderboard
2 Claude Fable 5.1 100% Sep 2026 Official leaderboard
3 Claude Opus 5.5 100% Sep 2026 Official leaderboard
4 Claude Sonnet 5.5 100% Sep 2026 Official leaderboard
5 Claude Fable 5.2 100% Nov 2026 Official leaderboard
6 Claude Opus 5 99% ±2 Jul 2026 Official leaderboard
7 GPT-6 Astra 99% ±2 Sep 2026 Official leaderboard
8 GPT-6.1 Sol 99% ±2 Sep 2026 Official leaderboard
9 Gemini 4 Argon 99% ±2 Sep 2026 Official leaderboard
10 Claude Fable 5 95% ±4.3 Jun 2026 Official leaderboard
11 Muse Wonder Full 94% ±4.7 Oct 2026 Official leaderboard
12 Kimi K3.1 Pro 91% ±5.6 Oct 2026 Official leaderboard
13 Claude Haiku 5.5 89% ±6.2 Oct 2026 Official leaderboard
14 Kimi K3 87% ±6.6 Jul 2026 Official leaderboard
15 Aristotle 86% ±6.8 Official leaderboard
16 Muse Wonder 84% ±7.2 Oct 2026 Official leaderboard
17 GPT-5.6 Sol 83% ±7.4 Jul 2026 Official leaderboard
18 GPT-6 Sol 83% ±7.4 Sep 2026 Official leaderboard
19 Claude Sonnet 5 77% ±8.3 Jun 2026 Official leaderboard
20 Tencent Hy4 preview 75% ±8.5 Aug 2026 Official leaderboard
21 GPT-5.6 Terra 74% ±8.6 Jul 2026 Official leaderboard
22 MiMo-V2.6 Pro 70% ±9 Sep 2026 Official leaderboard
23 GPT-6 Luna 64% ±9.5 Sep 2026 Official leaderboard
24 MiMo-V2.6 Flash 63% ±9.5 Sep 2026 Official leaderboard
25 Ember-1 61% ±9.6 Sep 2026 Official leaderboard
26 GPT-5.6 Luna 60% ±9.7 Jul 2026 Official leaderboard
27 Qwen 3.8 Max 58% ±9.7 Aug 2026 Official leaderboard
28 Gemini 3.7 Flash 58% ±9.7 Aug 2026 Official leaderboard
29 Muse Spark 1.3 58% ±9.7 max Sep 2026 Official leaderboard
30 DeepSeek V4 Flash 0731 56% ±9.8 Jul 2026 Official leaderboard
31 DeepSeek V4.1 Flash 54% ±9.8 Sep 2026 Official leaderboard
32 Grok 4.6 51% ±9.8 Aug 2026 Official leaderboard
33 DeepSeek V4 Pro 0813 50% ±9.8 Aug 2026 Official leaderboard
34 GLM-5.3 49% ±9.8 Aug 2026 Official leaderboard
35 Gemini 3.8 Flash 48% ±9.8 Sep 2026 Official leaderboard
36 Muse Spark 1.2 43% ±9.8 Aug 2026 Official leaderboard
37 Step 5 Preview 42% ±9.7 Sep 2026 Official leaderboard
38 Gemini 3.5 Flash 31% ±9.1 May 2026 Official leaderboard
39 Grok 4.5 31% ±9.1 Jul 2026 Official leaderboard
40 Gemini 3.1 Pro 26% ±8.6 Feb 2026 Official leaderboard
41 Grok 4.7 26% ±8.6 Sep 2026 Official leaderboard
42 MiMo-V2.5-Pro 22% ±8.2 Apr 2026 Official leaderboard
43 GLM-5.3-Flash 21% ±8 Aug 2026 Official leaderboard
44 MiniMax-M3 18% ±7.6 Jun 2026 Official leaderboard
45 Beam 1.0 18% ±7.6 Oct 2026 Official leaderboard
46 MiMo-V2.5 16% ±7.2 Apr 2026 Official leaderboard
47 DeepSeek-V4-Pro 16% ±7.2 Apr 2026 Official leaderboard
48 Qwen 3.8 27B 16% ±7.2 Aug 2026 Official leaderboard
49 Mistral Large 4 Preview 10% ±5.9 Oct 2026 Official leaderboard
50 Mistral Medium 3.5 9.00% ±5.6 Apr 2026 Official leaderboard
51 Inkling-Small 6.00% ±4.7 Jul 2026 Official leaderboard
52 Mercury 2.5 3.00% ±3.4 Sep 2026 Official leaderboard
53 Inkling 0% Jul 2026 Official leaderboard
54 Nemotron 3.5 Lightning 30B-A3B 0% Aug 2026 Official leaderboard
#ModelScore ± 95% CISettingDateSource
1 Claude Fable 5.1 100% Sep 2026 Official leaderboard
2 Claude Opus 5.5 100% Sep 2026 Official leaderboard
3 Claude Sonnet 5.5 100% Sep 2026 Official leaderboard
4 Claude Fable 5.2 100% Nov 2026 Official leaderboard
5 Claude Opus 5 99% ±2 Jul 2026 Official leaderboard
6 GPT-6 Astra 99% ±2 Sep 2026 Official leaderboard
7 GPT-6.1 Sol 99% ±2 Sep 2026 Official leaderboard
8 Gemini 4 Argon 99% ±2 Sep 2026 Official leaderboard
9 Claude Fable 5 95% ±4.3 Jun 2026 Official leaderboard
10 Muse Wonder Full 94% ±4.7 Oct 2026 Official leaderboard
11 Kimi K3.1 Pro 91% ±5.6 Oct 2026 Official leaderboard
12 Claude Haiku 5.5 89% ±6.2 Oct 2026 Official leaderboard
13 Kimi K3 87% ±6.6 Jul 2026 Official leaderboard
14 Muse Wonder 84% ±7.2 Oct 2026 Official leaderboard
15 GPT-5.6 Sol 83% ±7.4 Jul 2026 Official leaderboard
16 GPT-6 Sol 83% ±7.4 Sep 2026 Official leaderboard
17 Claude Sonnet 5 77% ±8.3 Jun 2026 Official leaderboard
18 Tencent Hy4 preview 75% ±8.5 Aug 2026 Official leaderboard
19 GPT-5.6 Terra 74% ±8.6 Jul 2026 Official leaderboard
20 MiMo-V2.6 Pro 70% ±9 Sep 2026 Official leaderboard
21 GPT-6 Luna 64% ±9.5 Sep 2026 Official leaderboard
22 MiMo-V2.6 Flash 63% ±9.5 Sep 2026 Official leaderboard
23 Ember-1 61% ±9.6 Sep 2026 Official leaderboard
24 GPT-5.6 Luna 60% ±9.7 Jul 2026 Official leaderboard
25 Qwen 3.8 Max 58% ±9.7 Aug 2026 Official leaderboard
26 Gemini 3.7 Flash 58% ±9.7 Aug 2026 Official leaderboard
27 Muse Spark 1.3 58% ±9.7 max Sep 2026 Official leaderboard
28 DeepSeek V4 Flash 0731 56% ±9.8 Jul 2026 Official leaderboard
29 DeepSeek V4.1 Flash 54% ±9.8 Sep 2026 Official leaderboard
30 Grok 4.6 51% ±9.8 Aug 2026 Official leaderboard
31 DeepSeek V4 Pro 0813 50% ±9.8 Aug 2026 Official leaderboard
32 GLM-5.3 49% ±9.8 Aug 2026 Official leaderboard
33 Gemini 3.8 Flash 48% ±9.8 Sep 2026 Official leaderboard
34 Muse Spark 1.2 43% ±9.8 Aug 2026 Official leaderboard
35 Step 5 Preview 42% ±9.7 Sep 2026 Official leaderboard
36 Gemini 3.5 Flash 31% ±9.1 May 2026 Official leaderboard
37 Grok 4.5 31% ±9.1 Jul 2026 Official leaderboard
38 Gemini 3.1 Pro 26% ±8.6 Feb 2026 Official leaderboard
39 Grok 4.7 26% ±8.6 Sep 2026 Official leaderboard
40 MiMo-V2.5-Pro 22% ±8.2 Apr 2026 Official leaderboard
41 GLM-5.3-Flash 21% ±8 Aug 2026 Official leaderboard
42 MiniMax-M3 18% ±7.6 Jun 2026 Official leaderboard
43 Beam 1.0 18% ±7.6 Oct 2026 Official leaderboard
44 MiMo-V2.5 16% ±7.2 Apr 2026 Official leaderboard
45 DeepSeek-V4-Pro 16% ±7.2 Apr 2026 Official leaderboard
46 Qwen 3.8 27B 16% ±7.2 Aug 2026 Official leaderboard
47 Mistral Large 4 Preview 10% ±5.9 Oct 2026 Official leaderboard
48 Mistral Medium 3.5 9.00% ±5.6 Apr 2026 Official leaderboard
49 Inkling-Small 6.00% ±4.7 Jul 2026 Official leaderboard
50 Mercury 2.5 3.00% ±3.4 Sep 2026 Official leaderboard
51 Inkling 0% Jul 2026 Official leaderboard
52 Nemotron 3.5 Lightning 30B-A3B 0% Aug 2026 Official leaderboard
#ModelScoreSettingDateSource
1 Claude Fable 5.1 100% Sep 2026 Official leaderboard
2 Claude Opus 5.5 100% Sep 2026 Official leaderboard
3 Claude Sonnet 5.5 100% Sep 2026 Official leaderboard
4 Claude Fable 5.2 100% Nov 2026 Official leaderboard
5 Claude Opus 5 99% Jul 2026 Official leaderboard
6 GPT-6 Astra 99% Sep 2026 Official leaderboard
7 GPT-6.1 Sol 99% Sep 2026 Official leaderboard
8 Gemini 4 Argon 99% Sep 2026 Official leaderboard
9 Claude Fable 5 95% Jun 2026 Official leaderboard
10 Muse Wonder Full 94% Oct 2026 Official leaderboard
11 Kimi K3.1 Pro 91% Oct 2026 Official leaderboard
12 Claude Haiku 5.5 89% Oct 2026 Official leaderboard
13 Kimi K3 87% Jul 2026 Official leaderboard
14 Muse Wonder 84% Oct 2026 Official leaderboard
15 GPT-5.6 Sol 83% Jul 2026 Official leaderboard
16 GPT-6 Sol 83% Sep 2026 Official leaderboard
17 Claude Sonnet 5 77% Jun 2026 Official leaderboard
18 Tencent Hy4 preview 75% Aug 2026 Official leaderboard
19 GPT-5.6 Terra 74% Jul 2026 Official leaderboard
20 MiMo-V2.6 Pro 70% Sep 2026 Official leaderboard
21 GPT-6 Luna 64% Sep 2026 Official leaderboard
22 MiMo-V2.6 Flash 63% Sep 2026 Official leaderboard
23 Ember-1 61% Sep 2026 Official leaderboard
24 GPT-5.6 Luna 60% Jul 2026 Official leaderboard
25 Qwen 3.8 Max 58% Aug 2026 Official leaderboard
26 Gemini 3.7 Flash 58% Aug 2026 Official leaderboard
27 Muse Spark 1.3 58% max Sep 2026 Official leaderboard
28 DeepSeek V4 Flash 0731 56% Jul 2026 Official leaderboard
29 DeepSeek V4.1 Flash 54% Sep 2026 Official leaderboard
30 Grok 4.6 51% Aug 2026 Official leaderboard
31 DeepSeek V4 Pro 0813 50% Aug 2026 Official leaderboard
32 GLM-5.3 49% Aug 2026 Official leaderboard
33 Gemini 3.8 Flash 48% Sep 2026 Official leaderboard
34 Muse Spark 1.2 43% Aug 2026 Official leaderboard
35 Step 5 Preview 42% Sep 2026 Official leaderboard
36 Gemini 3.5 Flash 31% May 2026 Official leaderboard
37 Grok 4.5 31% Jul 2026 Official leaderboard
38 Gemini 3.1 Pro 26% Feb 2026 Official leaderboard
39 Grok 4.7 26% Sep 2026 Official leaderboard
40 MiMo-V2.5-Pro 22% Apr 2026 Official leaderboard
41 GLM-5.3-Flash 21% Aug 2026 Official leaderboard
42 MiniMax-M3 18% Jun 2026 Official leaderboard
43 Beam 1.0 18% Oct 2026 Official leaderboard
44 Inkling 2 Mini 18% Nov 2026 Lab model card
45 MiMo-V2.5 16% Apr 2026 Official leaderboard
46 DeepSeek-V4-Pro 16% Apr 2026 Official leaderboard
47 Qwen 3.8 27B 16% Aug 2026 Official leaderboard
48 Mistral Large 4 Preview 10% Oct 2026 Official leaderboard
49 Mistral Medium 3.5 9.00% Apr 2026 Official leaderboard
50 Inkling-Small 6.00% Jul 2026 Official leaderboard
51 Mercury 2.5 3.00% Sep 2026 Official leaderboard
52 Inkling 0% Jul 2026 Official leaderboard
53 Nemotron 3.5 Lightning 30B-A3B 0% Aug 2026 Official leaderboard

Labs report scores for 1 model in their cards. Switch the scores above to rank them together.

Accuracy against latency

The Vals scatter chart with its x axis switched from Cost to Latency: mean time per test against accuracy, on a log scale. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 20% 40% 60% 80% 100% 1001,00010K Latency (mean per test) (s), log scale, lower is better Accuracy Claude Fable 5.2 0% 20% 40% 60% 80% 100% 1001,00010K Latency (mean per test) (s), log scale, lower is better Accuracy Claude Fable 5.2
ModelSettingAccuracyLatency (mean per test)
Claude Fable 5.2 100% 170.86 s
Claude Sonnet 5.5 100% 347.58 s
Claude Opus 5.5 100% 351.04 s
Claude Fable 5.1 100% 478.3 s
AlephProver 100% 1,108 s
GPT-6.1 Sol 99% 284.7 s
GPT-6 Astra 99% 307.87 s
Gemini 4 Argon 99% 380.25 s
Claude Opus 5 99% 590.89 s
Claude Fable 5 95% 822.74 s
Muse Wonder Full 94% 695.56 s
Kimi K3.1 Pro 91% 496.16 s
Claude Haiku 5.5 89% 418.05 s
Kimi K3 87% 2,071 s
Muse Wonder 84% 929.05 s
GPT-6 Sol 83% 304.51 s
GPT-5.6 Sol 83% 520.4 s
Claude Sonnet 5 77% 898.08 s
Tencent Hy4 preview 75% 1,374 s
GPT-5.6 Terra 74% 777.34 s
MiMo-V2.6 Pro 70% 3,694 s
GPT-6 Luna 64% 451.48 s
MiMo-V2.6 Flash 63% 3,770 s
Ember-1 61% 1,269 s
GPT-5.6 Luna 60% 513.69 s
Gemini 3.7 Flash 58% 423.7 s
Muse Spark 1.3 max 58% 568.7 s
Qwen 3.8 Max 58% 1,868 s
DeepSeek V4 Flash 0731 56% 933.98 s
Muse Spark 1.3 55% 534.06 s
DeepSeek V4.1 Flash 54% 546.6 s
Grok 4.6 51% 861.9 s
DeepSeek V4 Pro 0813 50% 732.37 s
GLM-5.3 49% 2,543 s
Gemini 3.8 Flash 48% 368.65 s
Muse Spark 1.2 43% 469.15 s
Step 5 Preview 42% 2,314 s
Gemini 3.5 Flash 31% 394.48 s
Grok 4.5 31% 668.86 s
Grok 4.7 26% 767.18 s
Gemini 3.1 Pro 26% 889.06 s
MiMo-V2.5-Pro 22% 1,177 s
GLM-5.3-Flash 21% 4,275 s
Beam 1.0 18% 266.66 s
MiniMax-M3 18% 718.66 s
DeepSeek-V4-Pro 16% 423.45 s
MiMo-V2.5 16% 1,256 s
Qwen 3.8 27B 16% 1,454 s
Mistral Large 4 Preview 10% 1,259 s
Mistral Medium 3.5 9% 479.05 s
Inkling-Small 6% 507.3 s
Mercury 2.5 3% 201.72 s
Inkling 0% 436.43 s
Nemotron 3.5 Lightning 30B-A3B 0% 919.25 s

Accuracy against latency

The Vals scatter chart with its x axis switched from Cost to Latency: mean time per test against accuracy, on a log scale. The line joins the runs that no other run beats on both measures. We redrew this chart from the numbers on the official leaderboard.

0% 20% 40% 60% 80% 100% 1001,00010K Latency (mean per test) (s), log scale, lower is better Accuracy Claude Fable 5.2 0% 20% 40% 60% 80% 100% 1001,00010K Latency (mean per test) (s), log scale, lower is better Accuracy Claude Fable 5.2
ModelSettingAccuracyLatency (mean per test)
Claude Fable 5.2 100% 170.86 s
Claude Sonnet 5.5 100% 347.58 s
Claude Opus 5.5 100% 351.04 s
Claude Fable 5.1 100% 478.3 s
GPT-6.1 Sol 99% 284.7 s
GPT-6 Astra 99% 307.87 s
Gemini 4 Argon 99% 380.25 s
Claude Opus 5 99% 590.89 s
Claude Fable 5 95% 822.74 s
Muse Wonder Full 94% 695.56 s
Kimi K3.1 Pro 91% 496.16 s
Claude Haiku 5.5 89% 418.05 s
Kimi K3 87% 2,071 s
Muse Wonder 84% 929.05 s
GPT-6 Sol 83% 304.51 s
GPT-5.6 Sol 83% 520.4 s
Claude Sonnet 5 77% 898.08 s
Tencent Hy4 preview 75% 1,374 s
GPT-5.6 Terra 74% 777.34 s
MiMo-V2.6 Pro 70% 3,694 s
GPT-6 Luna 64% 451.48 s
MiMo-V2.6 Flash 63% 3,770 s
Ember-1 61% 1,269 s
GPT-5.6 Luna 60% 513.69 s
Gemini 3.7 Flash 58% 423.7 s
Muse Spark 1.3 max 58% 568.7 s
Qwen 3.8 Max 58% 1,868 s
DeepSeek V4 Flash 0731 56% 933.98 s
Muse Spark 1.3 55% 534.06 s
DeepSeek V4.1 Flash 54% 546.6 s
Grok 4.6 51% 861.9 s
DeepSeek V4 Pro 0813 50% 732.37 s
GLM-5.3 49% 2,543 s
Gemini 3.8 Flash 48% 368.65 s
Muse Spark 1.2 43% 469.15 s
Step 5 Preview 42% 2,314 s
Gemini 3.5 Flash 31% 394.48 s
Grok 4.5 31% 668.86 s
Grok 4.7 26% 767.18 s
Gemini 3.1 Pro 26% 889.06 s
MiMo-V2.5-Pro 22% 1,177 s
GLM-5.3-Flash 21% 4,275 s
Beam 1.0 18% 266.66 s
MiniMax-M3 18% 718.66 s
DeepSeek-V4-Pro 16% 423.45 s
MiMo-V2.5 16% 1,256 s
Qwen 3.8 27B 16% 1,454 s
Mistral Large 4 Preview 10% 1,259 s
Mistral Medium 3.5 9% 479.05 s
Inkling-Small 6% 507.3 s
Mercury 2.5 3% 201.72 s
Inkling 0% 436.43 s
Nemotron 3.5 Lightning 30B-A3B 0% 919.25 s

The best score, 100%, costs $0.52 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6.1 Sol comes within 1 points of the best for $0.23 a task, about half the cost. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

NVIDIA OpenAI Anthropic Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Sonnet 5.5 GPT-6.1 Sol Claude Haiku 5.5 GPT-6 Luna Nemotron 3.5 Lightning 30B-A3B 0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Sonnet 5.5 Claude Haiku 5.5

Source: Official ProofBench v1.1 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Fable 5.1 100% $2.70 41,014
Claude Opus 5.5 100% $0.92 34,454
Claude Sonnet 5.5 100% $0.52 38,174
Claude Fable 5.2 100% $0.79 15,270
Claude Opus 5 99% $1.79 47,697
GPT-6 Astra 99% $1.71 9,972
GPT-6.1 Sol 99% $0.23 10,381
Gemini 4 Argon 99% $0.78 16,943
Claude Fable 5 95% $5.23 71,846
Muse Wonder Full 94% $0.86 27,090
Kimi K3.1 Pro 91% $0.47 22,381
Claude Haiku 5.5 89% $0.17 96,721
Kimi K3 87% $1.65 73,191
Muse Wonder 84% $1.31 36,876
GPT-5.6 Sol 83% $1.44 15,376
GPT-6 Sol 83% $0.48 13,630
Claude Sonnet 5 77% $1.37 75,841
Tencent Hy4 preview 75% $0.34 70,300
GPT-5.6 Terra 74% $1.27 15,580
MiMo-V2.6 Pro 70% $0.25 156,532
GPT-6 Luna 64% $0.04 26,528
MiMo-V2.6 Flash 63% $0.12 242,145
Ember-1 61% $1.74 72,478
GPT-5.6 Luna 60% $0.12 19,442
Qwen 3.8 Max 58% $1.50 109,428
Gemini 3.7 Flash 58% $0.56 13,973
Muse Spark 1.3 max 58% $0.48 45,292
DeepSeek V4 Flash 0731 56% $0.16 74,456
Muse Spark 1.3 55% $0.43 36,064
DeepSeek V4.1 Flash 54% $0.13 81,902
Grok 4.6 51% $0.82 38,271
DeepSeek V4 Pro 0813 50% $0.26 24,345
GLM-5.3 49% $2.08 195,563
Gemini 3.8 Flash 48% $0.60 12,467
Muse Spark 1.2 43% $0.44 38,717
Step 5 Preview 42% $0.71 133,184
Gemini 3.5 Flash 31% $0.57 14,092
Grok 4.5 31% $0.57 23,013
Gemini 3.1 Pro 26% $0.78 21,000
Grok 4.7 26% $0.79 14,226
MiMo-V2.5-Pro 22% $0.09 53,015
GLM-5.3-Flash 21% $0.45 333,212
MiniMax-M3 18% $0.42 48,763
Beam 1.0 18% 17,850
MiMo-V2.5 16% $0.06 119,561
DeepSeek-V4-Pro 16% $0.14 16,116
Qwen 3.8 27B 16% $2.30 169,025
Mistral Large 4 Preview 10% $0.33 18,237
Mistral Medium 3.5 9.00% $1.30 21,134
Inkling-Small 6.00% $0.08 25,970
Mercury 2.5 3.00% $0.04 15,608
Inkling 0% $0.20 14,174
Nemotron 3.5 Lightning 30B-A3B 0% $0.02 51,314

The best score, 100%, costs $0.52 a task

Runs on the cost frontier score higher than any run that's cheaper. GPT-6.1 Sol comes within 1 points of the best for $0.23 a task, about half the cost. The costs are the leaderboard's own estimates, so compare them only within this benchmark.

NVIDIA OpenAI Anthropic Cost frontier One model at each effort level
0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Sonnet 5.5 GPT-6.1 Sol Claude Haiku 5.5 GPT-6 Luna Nemotron 3.5 Lightning 30B-A3B 0% 20% 40% 60% 80% 100% $0.01 $0.10 $1.00 $10 Cost per task, log scale Claude Sonnet 5.5 Claude Haiku 5.5

Source: Official ProofBench v1.1 leaderboard

ModelEffortHarnessScoreCost per taskOutput tokens
Claude Fable 5.1 100% $2.70 41,014
Claude Opus 5.5 100% $0.92 34,454
Claude Sonnet 5.5 100% $0.52 38,174
Claude Fable 5.2 100% $0.79 15,270
Claude Opus 5 99% $1.79 47,697
GPT-6 Astra 99% $1.71 9,972
GPT-6.1 Sol 99% $0.23 10,381
Gemini 4 Argon 99% $0.78 16,943
Claude Fable 5 95% $5.23 71,846
Muse Wonder Full 94% $0.86 27,090
Kimi K3.1 Pro 91% $0.47 22,381
Claude Haiku 5.5 89% $0.17 96,721
Kimi K3 87% $1.65 73,191
Muse Wonder 84% $1.31 36,876
GPT-5.6 Sol 83% $1.44 15,376
GPT-6 Sol 83% $0.48 13,630
Claude Sonnet 5 77% $1.37 75,841
Tencent Hy4 preview 75% $0.34 70,300
GPT-5.6 Terra 74% $1.27 15,580
MiMo-V2.6 Pro 70% $0.25 156,532
GPT-6 Luna 64% $0.04 26,528
MiMo-V2.6 Flash 63% $0.12 242,145
Ember-1 61% $1.74 72,478
GPT-5.6 Luna 60% $0.12 19,442
Qwen 3.8 Max 58% $1.50 109,428
Gemini 3.7 Flash 58% $0.56 13,973
Muse Spark 1.3 max 58% $0.48 45,292
DeepSeek V4 Flash 0731 56% $0.16 74,456
Muse Spark 1.3 55% $0.43 36,064
DeepSeek V4.1 Flash 54% $0.13 81,902
Grok 4.6 51% $0.82 38,271
DeepSeek V4 Pro 0813 50% $0.26 24,345
GLM-5.3 49% $2.08 195,563
Gemini 3.8 Flash 48% $0.60 12,467
Muse Spark 1.2 43% $0.44 38,717
Step 5 Preview 42% $0.71 133,184
Gemini 3.5 Flash 31% $0.57 14,092
Grok 4.5 31% $0.57 23,013
Gemini 3.1 Pro 26% $0.78 21,000
Grok 4.7 26% $0.79 14,226
MiMo-V2.5-Pro 22% $0.09 53,015
GLM-5.3-Flash 21% $0.45 333,212
MiniMax-M3 18% $0.42 48,763
Beam 1.0 18% 17,850
MiMo-V2.5 16% $0.06 119,561
DeepSeek-V4-Pro 16% $0.14 16,116
Qwen 3.8 27B 16% $2.30 169,025
Mistral Large 4 Preview 10% $0.33 18,237
Mistral Medium 3.5 9.00% $1.30 21,134
Inkling-Small 6.00% $0.08 25,970
Mercury 2.5 3.00% $0.04 15,608
Inkling 0% $0.20 14,174
Nemotron 3.5 Lightning 30B-A3B 0% $0.02 51,314

2 models have held number one on the official leaderboard since its release in August 2026 2 models have held number one with claims included since its release in August 2026

Lab claims count from the day the card came out, independent runs from the model’s release.

Anthropic
Own leaderboard Passed 85% in Aug 2026, by Claude Opus 5

Source: Official ProofBench v1.1 leaderboard

Claude Fable 5 on the official leaderboard passed 85% in June 2026

Models released before it came out already scored that high. All the scores we have, one dot each, so a model tested at several settings appears more than once.

Plot with other benchmarks
On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 Claude Fable 5 Gemini 3.5 Flash Gemini 3.1 Pro 0% 20% 40% 60% 80% 100% 2026 2027 Near the ceiling (85%) Claude Fable 5.1 Gemini 3.5 Flash

Sources: Official ProofBench v1.1 leaderboard and labs’ model cards

Labs report it only in system cards

We read the headline results of 540 launch posts from 57 labs, and none of them list this benchmark. Some system cards do, below.

One system card reports it

Besides launch posts, we read the system cards, model cards and technical reports labs publish with a release. They report far more, including safety and frontier-risk tests.

DateDocumentScoreAs the document writes itSection
Nov 20, 2026 Inkling 2 Mini model card 18 ProofBench v1.1 Capabilities

Versions

We switch the top row to each new version when it comes out. Open it for one row per version. An old version stays in play while people still test it, so whoever first takes it past 85% gets credit for closing it out.

Anthropic OpenAI
ProofBench v1.1 Passed 85% in Aug 2026, by Claude Opus 5
ProofBench Replaced by v1.1, Aug 2026
ProofBench v1.1 Passed 85% in Aug 2026, by Claude Opus 5

Source: Official leaderboards

More in Math

All 140 in Math
  • AIME 2026 In use Models solve the 30 integer-answer problems from the 2026 AIME I and II math competitions. Scored by exact answer match. GPT-5.5 100%
  • IMO-AnswerBench In use 400 olympiad problems with short verifiable answers, robustified against memorization. Scored by answer match. GLM-5.2 91%
  • HMMT 2026 In use Models solve the 33 final-answer problems from the February 2026 Harvard-MIT Mathematics Tournament. Scored by exact answer match. GPT-5.5 98.5%
  • PolyMath In use Models solve 9,000 parallel math problems in 18 languages across four difficulty levels, from school math to frontier math. Final answers are checked. Qwen3.7-Max 86.5%
  • MathArena (Apex, IMO, USAMO) In use Evaluates models on newly released math competitions after release. Final-answer or human/LLM-judged proof scores. Claude Opus 4.8 81.3%
  • MathArena Apex Shortlist In use Models solve 47 hard final-answer problems that MathArena selected from 2025 competitions, where the best models scored about 50% at selection. Scored by accuracy over repeated attempts. GPT-5.5 98.4%

Data audit

All 10 checks pass. Every benchmark

Catalog record 2 pass
  • Pass Task access Public tasks, plus a private test set.
  • Pass Confidence Checked against a primary source.
Scores 4 pass 1 note
  • Pass Independent runs Scores from the official leaderboard (54 entries) and system cards (1 score).
  • Pass Models scored 53 models we track, plus 2 other entries on its leaderboard.
  • Pass Newest score The newest is from November 2026, for Inkling 2 Mini.
  • Pass Frontier models 7 of the top 10 have a score, led by Claude Fable 5.2 (#2).
  • Note Headroom Saturated. The best is 100%, past 85% since June 2026.
Leaderboard import 3 pass
  • Pass Importer We read the board on vals.ai: 53 rows for models we track and 2 other entries at the last read.
  • Pass Last read November 19th, 2026.
  • Pass Read worked The last read worked.
Consistency 1 pass 1 note
  • Note Lab claims No model has both a lab number and an independent run.
  • Pass Known issues No published audit found problems.

How we check