// BENCHMARK

Omni-MATH benchmark

AI model leaderboard for the Omni-MATH benchmark. Compare how large language models score on Omni-MATH, see the full ranking, and understand what this AI benchmark measures. DeepSeek-R1 currently leads with 85. Comprehensive olympiad-level mathematics benchmark covering 33 sub-domains and 10 difficulty tiers.

Leaderboard

# Model Organization Score Variant Source
#1 DeepSeek-R1 DeepSeek 85 cited-phi4r official ↗
#2 GPT-5 OpenAI 82.9 thinking-high-cited-ring-1t official ↗
#3 Ring-1T Ant Group 82.63 thinking official ↗
#4 Qwen3-235B-A22B-Thinking (Jul 2025) Qwen 82.52 thinking-cited-ring-1t official ↗
#5 gemini-2.5-pro Google DeepMind 82.14 cited-ring-1t official ↗
#6 DeepSeek-V3.1-Terminus DeepSeek 81.93 thinking-cited-ring-1t official ↗
#7 Phi-4-Reasoning-plus Microsoft 81.9 pass@1-avg@5 official ↗
#8 Phi-4-Reasoning Microsoft 76.6 pass@1-avg@5 official ↗
#9 o3-mini OpenAI 74.6 high-cited-phi4r official ↗
#10 Ling-1T Ant Group 74.46 official ↗
#11 gemini-2.5-pro Google DeepMind 72.02 lowthink-cited-ling-1t official ↗
#12 o1 OpenAI 67.5 cited-phi4r official ↗
#13 DeepSeek-V3.1-Terminus DeepSeek 64.77 non-thinking-cited-ling-1t official ↗
#14 DeepSeek-R1-Distill-Llama-70B DeepSeek 63.4 cited-phi4r official ↗
#15 Kimi K2 Moonshot 62.42 instruct-0905-cited-ling-1t official ↗
#16 GPT-5 OpenAI 61.09 main-cited-ling-1t official ↗
#17 o1-mini OpenAI 60.5 cited-phi4r official ↗
#18 Claude 3.7 Sonnet Anthropic 52.6 thinking-cited-phi4r official ↗

Frequently asked questions about Omni-MATH

What is the Omni-MATH benchmark?

Comprehensive olympiad-level mathematics benchmark covering 33 sub-domains and 10 difficulty tiers.

How is the Omni-MATH benchmark scored?

Omni-MATH is scored using the accuracy metric, where a higher score is better. GenAIList aggregates reported scores from model providers and papers into a single ranked leaderboard.

Which AI model scores highest on Omni-MATH?

As of the latest reported scores on GenAIList, DeepSeek-R1 achieves the highest result on Omni-MATH with a score of 85.

Is a higher Omni-MATH score better?

Yes. On Omni-MATH a higher score indicates better performance, so models near the top of the leaderboard are the strongest.