# MATH-500 leaderboard

Mathematics benchmark

> Every figure below is reproduced as its upstream source published it: nothing is modelled, estimated, interpolated or converted. `Unavailable` means no source published the value — it is never a zero. Each value carries its evidence state and the date it was observed.

Page: https://modelscale.dev/leaderboards/benchmarks/math-math500

MATH-500 Problem Set. Every model the catalog carries a published MATH-500 value for, ranked by that value.

## Benchmark record

| Field | Value |
| --- | --- |
| Category | Mathematics |
| Measure | Free-form mathematical answers |
| Tasks | 500 problems |
| Difficulty | High school to undergraduate |
| Published by | [Measuring Mathematical Problem Solving With the MATH Dataset](https://arxiv.org/abs/2103.03874) |

A curated subset of 500 problems from the MATH dataset, covering algebra, counting and probability, geometry, intermediate algebra, number theory, prealgebra, and precalculus.

## MATH-500 ranking

29 models with a published MATH-500 value, ordered by that value, highest first. Models the source has not scored on this benchmark are not listed — they are unmeasured, not last.

| Rank | Model | Provider | Free-form mathematical answers | Evidence |
| --- | --- | --- | --- | --- |
| 1 | [Ternary Bonsai 2 27B](https://modelscale.dev/models/ternary-bonsai-2-27b) | Prism ML | 98.8 | Observed 2026-11-21 · source benchlm:models |
| 2 | [OrcaCyber-Nano-1.0](https://modelscale.dev/models/orcacyber-nano-1-0) | OrcaRouter | 98.6 | Observed 2026-11-21 · source benchlm:models |
| 2 | [Sarvam 105B](https://modelscale.dev/models/sarvam-105b) | Sarvam | 98.6 | Observed 2026-11-21 · source benchlm:models |
| 4 | [Qwen3 235B-A22B (Reasoning)](https://modelscale.dev/models/qwen3-235b-a22b-reasoning) | Alibaba | 98 | Observed 2026-11-21 · source benchlm:models |
| 4 | [QwQ-32B](https://modelscale.dev/models/qwq-32b) | Alibaba | 98 | Observed 2026-11-21 · source benchlm:models |
| 6 | [o4-mini](https://modelscale.dev/models/o4-mini) | OpenAI | 97.5 | Observed 2026-11-21 · source benchlm:models |
| 7 | [Kimi K2](https://modelscale.dev/models/kimi-k2) | Moonshot AI | 97.4 | Observed 2026-11-21 · source benchlm:models |
| 8 | [DeepSeek-R1](https://modelscale.dev/models/deepseek-r1) | DeepSeek | 97.3 | Observed 2026-11-21 · source benchlm:models |
| 9 | [Qwen3 32B (Reasoning)](https://modelscale.dev/models/qwen3-32b-reasoning) | Alibaba | 97.2 | Observed 2026-11-21 · source benchlm:models |
| 10 | [Nemotron Ultra 253B](https://modelscale.dev/models/nemotron-ultra-253b) | NVIDIA | 97 | Observed 2026-11-21 · source benchlm:models |
| 10 | [Sarvam 30B](https://modelscale.dev/models/sarvam-30b) | Sarvam | 97 | Observed 2026-11-21 · source benchlm:models |
| 12 | [MiniMax M1 80k](https://modelscale.dev/models/minimax-m1-80k) | MiniMax | 96.8 | Observed 2026-11-21 · source benchlm:models |
| 13 | [Claude 3.7 Sonnet Thinking](https://modelscale.dev/models/claude-3-7-sonnet-thinking) | Anthropic | 96.2 | Observed 2026-11-21 · source benchlm:models |
| 14 | [LFM2.5-24B-A2B](https://modelscale.dev/models/lfm2-5-24b-a2b) | LiquidAI | 96.04 | Observed 2026-11-21 · source benchlm:models |
| 15 | [Magistral Small](https://modelscale.dev/models/magistral-small) | Mistral | 95.9 | Observed 2026-11-21 · source benchlm:models |
| 16 | [LongCat-Flash-Lite-Sparse](https://modelscale.dev/models/longcat-flash-lite-sparse) | Meituan | 95.8 | Observed 2026-11-21 · source benchlm:models |
| 17 | [MiniCPM5-2B](https://modelscale.dev/models/minicpm5-2b) | OpenBMB | 94.6 | Observed 2026-11-21 · source benchlm:models |
| 18 | [DeepSeek R1 Distill Llama 70B](https://modelscale.dev/models/deepseek-r1-distill-llama-70b) | DeepSeek | 94.5 | Observed 2026-11-21 · source benchlm:models |
| 19 | [Magistral Medium](https://modelscale.dev/models/magistral-medium) | Mistral | 94.3 | Observed 2026-11-21 · source benchlm:models |
| 20 | [DeepSeek R1 Distill Qwen 14B](https://modelscale.dev/models/deepseek-r1-distill-qwen-14b) | DeepSeek | 93.9 | Observed 2026-11-21 · source benchlm:models |
| 21 | [MiniCPM5-1B](https://modelscale.dev/models/minicpm5-1b) | OpenBMB | 91.6 | Observed 2026-11-21 · source benchlm:models |
| 22 | [Qwen3 235B-A22B](https://modelscale.dev/models/qwen3-235b-a22b) | Alibaba | 91.2 | Observed 2026-11-21 · source benchlm:models |
| 23 | [DeepSeek V3](https://modelscale.dev/models/deepseek-v3) | DeepSeek | 90.2 | Observed 2026-11-21 · source benchlm:models |
| 24 | [o1-mini](https://modelscale.dev/models/o1-mini) | OpenAI | 90 | Observed 2026-11-21 · source benchlm:models |
| 25 | [LFM2.5-8B-A1B](https://modelscale.dev/models/lfm2-5-8b-a1b) | LiquidAI | 88.76 | Observed 2026-11-21 · source benchlm:models |
| 26 | [Qwen3 32B](https://modelscale.dev/models/qwen3-32b) | Alibaba | 88.6 | Observed 2026-11-21 · source benchlm:models |
| 27 | [Claude 3.7 Sonnet](https://modelscale.dev/models/claude-3-7-sonnet) | Anthropic | 82.2 | Observed 2026-11-21 · source benchlm:models |
| 28 | [Kanana-2 1.3B Instruct](https://modelscale.dev/models/kanana-2-1-3b-instruct) | Kakao | 61.4 | Observed 2026-11-21 · source benchlm:models |
| 29 | [Kanana-2 3B Instruct](https://modelscale.dev/models/kanana-2-3b-instruct) | Kakao | 61.2 | Observed 2026-11-21 · source benchlm:models |

Rows are ordered by the value the source published, highest first. The source does not state whether a higher value is the better result, so this document does not either: for a benchmark that measures a rate of failure, read the table from the bottom.
