math benchmark
IMO-AnswerBench Leaderboard
IMO-AnswerBench evaluates mathematical reasoning on International Mathematical Olympiad (IMO) problems, focusing on answer generation and verification.
Updated Sep 5, 2026
Models20
Model coverage20
MetricScore
EvidenceB
IMO-AnswerBench Ranking
Higher score ranks better on this benchmark.
Sort by
Rank | Model | Score | Percentile | Participants | Evidence | Evaluated |
|---|
| Rank01 | ModelNV | Score92.30% | Percentile100.00% | Participants20 | EvidenceC | Evaluated |
| Rank02 | ModelZA | Score91.00% | Percentile94.74% | Participants20 | EvidenceC | Evaluated |
| Rank03 | ModelTE | Score90.00% | Percentile89.47% | Participants20 | EvidenceC | Evaluated |
| Rank04 | ModelAC | Score90.00% | Percentile84.21% | Participants20 | EvidenceC | Evaluated |
| Rank05 | ModelDE | Score89.80% | Percentile78.95% | Participants20 | EvidenceC | Evaluated |
| Rank06 | ModelDE | Score88.40% | Percentile73.68% | Participants20 | EvidenceC | Evaluated |
| Rank07 | ModelMA | Score86.00% | Percentile68.42% | Participants20 | EvidenceC | Evaluated |
| Rank08 | ModelAC | Score86.00% | Percentile63.16% | Participants20 | EvidenceC | Evaluated |
| Rank09 | ModelST | Score85.40% | Percentile57.89% | Participants20 | EvidenceC | Evaluated |
| Rank10 | ModelDE | Score85.10% | Percentile52.63% | Participants20 | EvidenceC | Evaluated |
| Rank11 | ModelZA | Score83.80% | Percentile47.37% | Participants20 | EvidenceC | Evaluated |
| Rank12 | ModelAC | Score83.80% | Percentile42.11% | Participants20 | EvidenceC | Evaluated |
| Rank13 | ModelZA | Score82.00% | Percentile36.84% | Participants20 | EvidenceC | Evaluated |
| Rank14 | ModelMA | Score81.80% | Percentile31.58% | Participants20 | EvidenceC | Evaluated |
| Rank15 | ModelAC | Score80.90% | Percentile26.32% | Participants20 | EvidenceC | Evaluated |
| Rank16 | ModelAC | Score80.80% | Percentile21.05% | Participants20 | EvidenceC | Evaluated |
| Rank17 | ModelAC | Score78.90% | Percentile15.79% | Participants20 | EvidenceC | Evaluated |
| Rank18 | ModelMA | Score78.60% | Percentile10.53% | Participants20 | EvidenceC | Evaluated |
| Rank19 | ModelME | Score78.60% | Percentile5.26% | Participants20 | EvidenceC | Evaluated |
| Rank20 | ModelDE | Score78.30% | Percentile0.00% | Participants20 | EvidenceC | Evaluated |
IMO-AnswerBench Highlights
The leading models and scores on this benchmark.
IMO-AnswerBench Score Distribution
A closer view of the leading scores on this benchmark.
The Top AI Models for IMO-AnswerBench
The first five results on this benchmark, with official price and output speed added where the model identity can be matched.
What is IMO-AnswerBench?
What IMO-AnswerBench measures and how its scores work.
IMO-AnswerBench is a benchmark for evaluating mathematical reasoning capabilities on International Mathematical Olympiad (IMO) problems.
It measures answer generation and verification on IMO problems using Score, reported as a ratio.
Scores are shown in ratio. This benchmark is not independently verified and has an evidence level of B.
- Family
- IMO-AnswerBench
- Modality
- text
- Primary category
- math
- Score direction
- higher
- LLMBoard eligible
- No
- Evaluation key
- overall
LLM Stats. Benchmark scores retain their original unit. Overall score eligibility is shown separately.
FAQ
Common questions about IMO-AnswerBench.
Which model scores highest on IMO-AnswerBench?
Nemotron 3 Ultra (550B A55B) is currently ranked first with 92.30%.
What are the top three models on IMO-AnswerBench?
The current leaders are Nemotron 3 Ultra (550B A55B) (92.30%), GLM-5.2 (91.00%), and Hy3 (90.00%).
Which IMO-AnswerBench model has the lowest official input price?
Step-3.5-Flash has the lowest matched official input price at $0.10 input / $0.30 output per 1M tokens.
Which models are fastest among IMO-AnswerBench results?
The fastest matched records are DeepSeek-V3.2 (97.00 tok/s via DeepInfra), Step-3.5-Flash (21.16 tok/s via StepFun), and GLM-5.2 (3.96 tok/s via ZAI).
Does the highest score result prove overall model quality?
No. This benchmark measures one defined capability or task. The overall LLMBoard score uses a separate aggregation across eligible benchmark evidence.