Skip to main content

Benchmark profile

Vals ProofBench (ProofBench)

Vals AI automated theorem-proving benchmark.

Data verified

How BenchLM shows ProofBench

BenchLM mirrors the public Vals AI ProofBench leaderboard captured from https://www.vals.ai/benchmarks/proof_bench and updated by Vals on July 23, 2026. The snapshot preserves overall scores, uncertainty, latency, cost-per-test metadata, and task-level scores where Vals publishes them.

ProofBench is display only on BenchLM. Vals proprietary or Vals-hosted aggregate views are useful context, but BenchLM does not use them as weighted ranking inputs or as a replacement for benchmark-native source records.

54 Vals rows1 task viewsprivate datasetTasks: OverallDisplay only

ProofBench score on ProofBench — July 23, 2026

BenchLM mirrors the published proofbench score view for ProofBench. Claude Opus 5 leads the public snapshot at 78.00% , followed by GPT-5.6 Sol (77.00%) and Claude Fable 5 (77.00%). BenchLM does not use these results to rank models overall.

54 modelsExternal benchmark mirrorsCurrentDisplay onlyUpdated July 23, 2026

ProofBench score table (54 models)

Score
1
Claude Opus 5Anthropic
78.00%
2
77.00%
3
77.00%
4
71.00%
5
AristotleHarmonic
71.00%
6
Kimi K3Moonshot AI
70.00%
7
69.00%
8
66.00%
9
GPT-5.4OpenAI
56.00%
10
54.00%
11
54.00%
12
GPT-5.5OpenAI
50.00%
13
50.00%
14
45.00%
15
39.00%
16
36.00%
18
GLM 5.2Zhipu AI
35.00%
19
30.00%
20
29.00%
21
26.00%
23
24.00%
24
GLM 5.1Zhipu AI
22.22%
25
21.00%
26
20.00%
27
MiniMax M3MiniMax
19.00%
29
GPT-5OpenAI
18.00%
30
17.00%
31
Kimi K2.6Moonshot AI
16.00%
32
Mimo V2.5Xiaomi
16.00%
34
GPT-5.2OpenAI
15.00%
36
13.00%
37
12.00%
38
11.00%
39
10.00%
40
10.00%
41
9.00%
42
9.00%
43
8.00%
44
DeepSeek V3p2Fireworks AI
8.00%
45
GLM 4.7Zhipu AI
6.00%
46
5.00%
48
4.00%
49
4.00%
50
3.00%
51
InklingThinkingmachines
2.00%
53
Laguna Xs.2Poolside
0.00%
54
Laguna M.1Poolside
0.00%

The published ProofBench snapshot places Claude Opus 5 first at 78.00%. The third row is 1.00 points behind. The broader top-10 range is 24.00 points, so the table still separates the published systems.

54 models have been evaluated on ProofBench. The benchmark falls in the External benchmark mirrors category. We keep external benchmark mirrors separate from the weighted global scoring system, so these results remain source-specific evidence. ProofBench is currently displayed for reference but excluded from the scoring formula, so it does not directly affect overall rankings.

About ProofBench

Year

2026

Tasks

Automated theorem proving

Format

Accuracy score

Difficulty

Formal proof reasoning

BenchLM mirrors Vals ProofBench as a display-only math and proof benchmark.

BenchLM freshness & provenance

Version

ProofBench 2026

Refresh cadence

Quarterly

Staleness state

Current

Question availability

Public benchmark set

CurrentDisplay only

BenchLM uses freshness metadata to decide whether a benchmark should still be treated as a strong differentiator, a benchmark to watch, or a display-only reference. For the full scoring policy, see the BenchLM methodology page.

FAQ

What does ProofBench measure?

Vals AI automated theorem-proving benchmark.

Which model leads the published ProofBench snapshot?

Claude Opus 5 currently leads the published ProofBench snapshot with 78.00% proofbench score. BenchLM shows this benchmark for display only and does not use it in overall rankings.

How many models are evaluated on ProofBench?

54 AI models are included in BenchLM's mirrored ProofBench snapshot, based on the public leaderboard captured on July 23, 2026.

Last updated: July 23, 2026 · mirrored from the public benchmark leaderboard

Know when it’s worth switching models

The model to choose, the cheaper alternative, and the release we would wait on.

Read a sample issue

Join 2,000+ readers.

One email each week. Unsubscribe anytime.