Benchmark Almanac

IMO-ProofBench

Aletheia 91.9% tops its leaderboard GPT-5.5 Pro 88.1% leads the models we track Fading

Models write proofs for 60 olympiad problems, split into a basic set and an advanced set up to IMO-hard. Each proof is graded from 0 to 7 by experts or a Gemini autograder.

Part of IMO-Bench (EMNLP 2025). The v2 file only fixed a typo. In May 2026 Aletheia led the advanced set at 91.9 percent and GPT-5.5 Pro scored 88.1 percent.

In the authors’ words

Finding the right north-star metrics is highly critical for advancing the mathematical reasoning capabilities of foundation models, especially given that existing evaluations are either too easy or only focus on getting correct short answers. To address these issues, we present IMO-Bench, a suite of advanced reasoning benchmarks, vetted by a panel of top specialists and that specifically targets the level of the International Mathematical Olympiad (IMO), the most prestigious venue for young mathematicians. IMO-AnswerBench first tests models on 400 diverse Olympiad problems with verifiable short answers. IMO-Proof Bench is the next-level evaluation for proof-writing capabilities, which includes both basic and advanced IMO level problems as well as detailed grading guidelines to facilitate automatic grading. These benchmarks played a crucial role in our historic achievement of the gold-level performance at IMO 2025 with Gemini Deep Think (Luong and Lockhart, 2025). Our model achieved 80.0% on IMO-AnswerBench and 65.7% on the advanced IMO-Proof Bench, surpassing the best non-Gemini models by large margins of 6.9% and 42.4% respectively. We also showed that autograders built with Gemini reasoning correlate well with human evaluations and construct IMO-GradingBench, with 1000 human gradings on proofs, to enable further progress in automatic evaluation of long-form answers. We hope that IMO-Bench will help the community towards advancing robust mathematical reasoning and release it at https://imobench.github.io/.

Abstract of Towards Robust Mathematical Reasoning (arXiv:2511.01846, CC0)
Made by Google DeepMind (Superhuman Reasoning team) Released Nov 2025 Written proofs All tasks public Last in a launch post Sep 2026 , dropped after it saturated Last model added Sep 2026 Scores refreshed 2 Nov 2026
Level
Competition The strongest students, in olympiad-style contests
Scored
Graded by people Human experts grade or compare the outputs.
Answers
Written proof
google-deepmind/superhuman 807 stars Paper Dataset Leaderboard imobench.github.io
12
Entries on its own leaderboard, 10 models we track
3
Launch posts list it, from 3 labs
4
System cards report it, from 3 labs

Aletheia tops the leaderboard with 91.9%

Best score per entry from the benchmark's own leaderboard, read on 2 Nov 2026. Where a model ran with several settings or harnesses, we keep its best. Names in gray aren't in our model list: fine-tunes, agent systems, products and older models. They don't count toward leaders. The best model we track is GPT-5.5 Pro, at 88.1%.

The newest model with a score here came out in Apr 2026, and nobody has tested a newer one. Among the 20 highest-ranked models on the capability index, these have no score yet:

  1. 91.9 Aletheia
  2. 88.1 GPT-5.5 Pro
  3. 76.7 Gemini 3 Deep Think
  4. 71.9 GPT-5.5
  5. 65.7 Gemini Deep Think (IMO Gold)
  6. 49.0 Gemini 3.1 Pro
  7. 35.7 GPT-5.2
  8. 30.0 Gemini 3 Pro
  9. 28.6 GPT-5 Pro
  10. 23.8 Claude Opus 4.5
  11. 18.6 Grok 4.1 Fast
  12. 7.10 GPT-5.1
  1. 1 Aletheia 91.9%
  2. 2 GPT-5.5 Pro xhigh 88.1%
  3. 3 Gemini 3 Deep Think 76.7%
  4. 4 GPT-5.5 xhigh 71.9%
  5. 5 Gemini Deep Think (IMO Gold) 65.7%
  6. 6 Gemini 3.1 Pro 49%
  7. 7 GPT-5.2 high 35.7%
  8. 8 Gemini 3 Pro 30%
  9. 9 GPT-5 Pro 28.6%
  10. 10 Claude Opus 4.5 23.8%
  11. 11 Grok 4.1 Fast reasoning 18.6%
  12. 12 GPT-5.1 7.10%
#EntryScoreSettingDateSource
1 Aletheia 91.9% Feb 2026 Official leaderboard
2 GPT-5.5 Pro 88.1% xhigh Apr 2026 Official leaderboard
3 Gemini 3 Deep Think 76.7% Feb 2026 Official leaderboard
4 GPT-5.5 71.9% xhigh Apr 2026 Official leaderboard
5 Gemini Deep Think (IMO Gold) 65.7% Aug 2025 Official leaderboard
6 Gemini 3.1 Pro 49% Feb 2026 Official leaderboard
7 GPT-5.2 35.7% high Dec 2025 Official leaderboard
8 Gemini 3 Pro 30% Nov 2025 Official leaderboard
9 GPT-5 Pro 28.6% Oct 2025 Official leaderboard
10 Claude Opus 4.5 23.8% Nov 2025 Official leaderboard
11 Grok 4.1 Fast 18.6% reasoning Nov 2025 Official leaderboard
12 GPT-5.1 7.10% Nov 2025 Official leaderboard

Labs report scores for 5 models in their cards. We plot every source in the charts below.

5 models have held number one since its release in Nov 2025

Lab claims count from the day the card came out, independent runs from the model’s release. We draw one row each for the benchmark’s own leaderboard and what labs claim in their cards.

OpenAI Google DeepMind DeepSeek NVIDIA
Own leaderboard Passed 85% in Apr 2026, by GPT-5.5 Pro
GPT-5.5 Pro 88.1%
Lab claims Not tested since Sep 2026
Nemotron 3 Ultra 82.3%

The best score passed 85% in Apr 2026

All the scores we have, one dot each, so a model tested at several settings appears more than once.

On the benchmark’s own leaderboard Reported by the lab in its model card Best so far
0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.5 Pro Gemini 3 Deep Think DeepSeekMath-V2 Gemini 3 Pro 0% 20% 40% 60% 80% 100% 2026 Near the ceiling (85%) GPT-5.5 Pro Gemini 3 Deep Think

Labs headline it less often now, last in Sep 2026

Launch posts from every lab that has listed it, one tick per post. Tall colored ticks include it, and short gray ones leave it out.

DeepSeek
NVIDIA
Shanghai AI Lab
DateLaunch postScoreAs the post writes it
13 Sep 2026 Intern-S2-397B 38.4 IMO-Proof (Olympiad-Level Mathematics Proof)
19 Mar 2026 Nemotron-Cascade 2 30B-A3B 72.9 IMO ProofBench
27 Nov 2025 DeepSeekMath-V2 61.9 ProofBench-Advanced, DeepSeekMath-V2 (Heavy), human evaluations (Basic 99.0)

4 system cards report it, from 3 labs

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
23 Jun 2026 Seed2.1 Pro model card 54.3 IMOProof-Adv, Seed2.1 Pro column (Table 7). Deep Think 83.8 Capabilities
9 Jun 2026 Nemotron 3 Ultra technical report 82.3 IMO-ProofBench Advanced, 173/210, generate-verify-refine test-time scaling, human graded (Table 11) Capabilities
19 Mar 2026 Nemotron-Cascade 2 30B-A3B technical report 72.9 IMO Proof Bench Overall (60), generate-verify-refine test-time scaling, DeepSeek-V3.2-Speciale judge (Tables 1 and 5). Basic 92.5, Advanced 53.4 Capabilities
27 Nov 2025 DeepSeekMath-V2 technical report 61.9 ProofBench-Advanced, DeepSeekMath-V2 (Heavy), expert evaluation (Figure 3). ProofBench-Basic 99.0 Capabilities

More in Math

All 105 in Math
  • IMO-AnswerBench In use 400 olympiad problems with short verifiable answers, robustified against memorization. Scored by answer match. GLM-5.2 91%
  • 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%
  • 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%
  • 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%
  • 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%
  • USAMO 2026 In use Models write full proofs for the six problems of the 2026 USA Mathematical Olympiad. Judges grade each proof from 0 to 7 against rubrics. GPT-5.5 98.2%