MATH level 5
Each model’s result as Epoch AI’s Benchmarking Hub publishes it, run by Epoch or collected from the source it cites. Facthory has not run them.
Data: Epoch AI, ‘Capabilities & benchmarking’, epoch.ai/benchmarks, licensed CC BY 4.0. Read on 16 Nov 2026.
The chart shows the 25 best and the table the 100 best of 108 results.
| # | Model | Organization | Result | Evidence |
|---|---|---|---|---|
| 1 | GPT-5 (high) API access | OpenAI | 98.1% ± 0.3% | – |
| 2 | GPT-5 (medium) API access | OpenAI | 97.9% ± 0.3% | – |
| 3 | GPT-5 mini (high) API access | OpenAI | 97.8% ± 0.3% | – |
| 4 | o4-mini (high) API access | OpenAI | 97.8% ± 0.3% | – |
| 5 | o3 (high) API access | OpenAI | 97.8% ± 0.3% | – |
| 6 | Claude Sonnet 4.5 (32k thinking) API access | Anthropic | 97.7% ± 0.4% | – |
| 7 | Qwen3-Max-Instruct API access | Alibaba | 97.1% ± 0.4% | – |
| 8 | GPT-5 mini (medium) API access | OpenAI | 96.8% ± 0.4% | – |
| 9 | DeepSeek-R1 (May 2025) Open weights (unrestricted) | DeepSeek | 96.6% ± 0.4% | – |
| 10 | o3-mini (high) API access | OpenAI | 96.5% ± 0.4% | – |
| 11 | claude-haiku-4-5-20251001_32K API access | Anthropic | 96.4% ± 0.5% | – |
| 12 | Gemini 2.5 Pro Preview (Jun 2025) API access | Google DeepMind | 95.9% ± 0.4% | – |
| 13 | Gemini 2.5 Pro Preview (Mar 2025) API access | Google DeepMind | 95.6% ± 0.4% | – |
| 14 | GPT-5 nano (medium) API access | OpenAI | 95.2% ± 0.5% | – |
| 15 | o3-mini (medium) API access | OpenAI | 95.2% ± 0.5% | – |
| 16 | GPT-5 nano (high) API access | OpenAI | 94.9% ± 0.5% | – |
| 17 | o1 (high) API access | OpenAI | 94.7% ± 0.6% | – |
| 18 | o1 (medium) API access | OpenAI | 94.4% ± 0.6% | – |
| 19 | DeepSeek-R1 Open weights (unrestricted) | DeepSeek | 93.1% ± 0.7% | – |
| 20 | Claude 3.7 Sonnet (64k thinking) API access | Anthropic | 91.2% ± 0.8% | – |
| 21 | grok-3-mini-beta_low API access | xAI | 90.9% ± 0.8% | – |
| 22 | Claude 3.7 Sonnet (32k thinking) API access | Anthropic | 90% ± 0.8% | – |
| 23 | DeepSeek-R1-Distill-Llama-70B Open weights (unrestricted) | DeepSeek | 89.9% ± 0.7% | – |
| 24 | o1-mini (high) API access | OpenAI | 89.2% ± 0.7% | – |
| 25 | grok-3-beta API access | xAI | 88.7% ± 0.9% | – |
| 26 | grok-3-mini-beta_high API access | xAI | 88.1% ± 0.9% | – |
| 27 | GPT-4.1 mini API access | OpenAI | 87.3% ± 0.7% | – |
| 28 | DeepSeek-R1-Distill-Qwen-14B Open weights (unrestricted) | DeepSeek | 87.1% ± 0.7% | – |
| 29 | claude-haiku-4-5-20251001 API access | Anthropic | 86.9% ± 0.8% | – |
| 30 | Claude 3.7 Sonnet (16k thinking) API access | Anthropic | 86.3% ± 0.9% | – |
| 31 | Claude Opus 4 API access | Anthropic | 85% ± 1% | – |
| 32 | claude-sonnet-4-20250514 API access | Anthropic | 84.4% ± 1% | – |
| 33 | o1-mini (medium) API access | OpenAI | 84.3% ± 0.8% | – |
| 34 | Gemini 2.0 Pro Exp (Feb 2025) Hosted access (no API) | Google DeepMind | 83.5% ± 1% | – |
| 35 | GPT-4.1 API access | OpenAI | 83% ± 0.8% | – |
| 36 | Gemini 2.0 Flash (Feb 2025) API access | Google DeepMind,Google | 82.2% ± 0.9% | – |
| 37 | o1-preview API access | OpenAI | 81.6% ± 1.1% | – |
| 38 | mistral-medium-2505 API access | Mistral AI | 81.6% ± 0.9% | – |
| 39 | GPT-4.5 Preview (Feb 2025) API access | OpenAI | 78.6% ± 1.1% | – |
| 40 | DeepSeek-V3 (Mar 2025) Open weights (restricted use) | DeepSeek | 75.5% ± 0.9% | – |
| 41 | gemma-3-27b-it Open weights (restricted use) | Google DeepMind | 74% ± 1.1% | – |
| 42 | Llama 4 Maverick (FP8) Open weights (restricted use) | Meta AI | 73% ± 1.1% | – |
| 43 | gemini-1.5-pro-002 API access | Google DeepMind | 70.4% ± 1% | – |
| 44 | GPT-4.1 nano API access | OpenAI | 70% ± 1% | – |
| 45 | qwen3-235b-a22b Open weights (unrestricted) | Alibaba | 68.9% ± 0.9% | – |
| 46 | claude-3-7-sonnet-20250219 API access | Anthropic | 68.2% ± 1.1% | – |
| 47 | qwen-max-2025-01-25 API access | Alibaba | 67.2% ± 1.1% | – |
| 48 | qwen-plus-2025-01-25 API access | Alibaba | 65.3% ± 1.1% | – |
| 49 | phi-4 Open weights (unrestricted) | Microsoft Research | 64.9% ± 1.1% | – |
| 50 | DeepSeek-V3 Open weights (restricted use) | DeepSeek | 64.9% ± 1% | – |
| 51 | grok-2-1212 API access | xAI | 63.5% ± 1.3% | – |
| 52 | qwen2.5-72b-instruct Open weights (unrestricted) | Alibaba | 63.2% ± 1.1% | – |
| 53 | Llama-4-Scout-17B-16E-Instruct Open weights (restricted use) | Meta AI | 62.3% ± 1.2% | – |
| 54 | gemini-1.5-flash-002 API access | Google DeepMind | 61.9% ± 1.1% | – |
| 55 | Claude 3.5 Sonnet (Oct 2024) API access | Anthropic | 56.9% ± 1.1% | – |
| 56 | qwen-turbo-2024-11-01 API access | Alibaba | 56.2% ± 1.1% | – |
| 57 | qwen2.5-32b-instruct Open weights (unrestricted) | Alibaba | 56.1% ± 1.1% | – |
| 58 | GPT-4o (Aug 2024) API access | OpenAI | 53.3% ± 1.1% | – |
| 59 | gpt-4o-mini-2024-07-18 API access | OpenAI | 52.6% ± 1.1% | – |
| 60 | Claude 3.5 Sonnet (Jun 2024) API access | Anthropic | 51.7% ± 1.2% | – |
| 61 | GPT-4o (May 2024) API access | OpenAI | 51% ± 1.1% | – |
| 62 | mistral-large-2411 Open weights (non-commercial) | Mistral AI | 50.3% ± 1.1% | – |
| 63 | Llama-3.1-405B-Instruct Open weights (restricted use) | Meta AI | 49.8% ± 1.2% | – |
| 64 | GPT-4o (Nov 2024) API access | OpenAI | 49.8% ± 1.1% | – |
| 65 | mistral-small-2503 Open weights (unrestricted) | Mistral AI | 46.8% ± 1.2% | – |
| 66 | gpt-4-turbo-2024-04-09 API access | OpenAI | 46.7% ± 1.1% | – |
| 67 | Claude 3.5 Haiku (Oct 2024) API access | Anthropic | 46.4% ± 1.2% | – |
| 68 | mistral-large-2407 Open weights (non-commercial) | Mistral AI | 44.8% ± 1.1% | – |
| 69 | mistral-small-2501 Open weights (unrestricted) | Mistral AI | 44.8% ± 1.1% | – |
| 70 | Llama-3.1-Tulu-3-70B-DPO Open weights (restricted use) | Allen Institute for AI,University of Washington | 42.7% ± 1% | – |
| 71 | Llama-3.3-70B-Instruct Open weights (restricted use) | Meta AI | 41.6% ± 1.2% | – |
| 72 | gemini-1.5-pro-001 API access | Google DeepMind | 40.7% ± 1.1% | – |
| 73 | GPT-4 Turbo Preview (Nov 2023) API access | OpenAI | 40% ± 1.1% | – |
| 74 | Llama-3.2-90B-Vision-Instruct Open weights (restricted use) | Meta AI | 39.4% ± 1% | – |
| 75 | qwen2-72b-instruct Open weights (unrestricted) | Alibaba | 39.1% ± 1% | – |
| 76 | claude-3-opus-20240229 API access | Anthropic | 37.5% ± 1.1% | – |
| 77 | Llama-3.1-70B-Instruct Open weights (restricted use) | Meta AI | 36.7% ± 1% | – |
| 78 | GPT-4 Turbo Preview (January 2024) API access | OpenAI | 35.4% ± 1.1% | – |
| 79 | gemma-2-27b-it Open weights (restricted use) | Google DeepMind | 27.9% ± 1% | – |
| 80 | WizardLM-2-8x22B Open weights (unrestricted) | Microsoft | 25.7% ± 0.9% | – |
| 81 | Yi-1.5-34B-Chat Open weights (restricted use) | 01.AI | 25.5% ± 0.9% | – |
| 82 | Gemini 1.5 Flash (May 2024) API access | Google DeepMind | 25.1% ± 0.9% | – |
| 83 | mistral-large-2402 API access | Mistral AI | 24.5% ± 0.9% | – |
| 84 | open-mixtral-8x22b Open weights (unrestricted) | Mistral AI | 24.2% ± 0.8% | – |
| 85 | GPT-4 (Jun 2023) API access | OpenAI | 23% ± 0.8% | – |
| 86 | Llama-3.1-8B-Instruct Open weights (restricted use) | Meta AI | 22.9% ± 0.9% | – |
| 87 | Hermes-2-Theta-Llama-3-70B Open weights (restricted use) | Nous Research,Arcee AI | 22.7% ± 0.8% | – |
| 88 | Meta-Llama-3-70B-Instruct Open weights (restricted use) | Meta AI | 22.6% ± 0.8% | – |
| 89 | gemma-2-9b-it Open weights (restricted use) | Google DeepMind | 21% ± 0.9% | – |
| 90 | claude-3-sonnet-20240229 API access | Anthropic | 18.2% ± 0.8% | – |
| 91 | Phi-3-medium-128k-instruct Open weights (unrestricted) | Microsoft | 17.6% ± 0.7% | – |
| 92 | GPT-3.5 Turbo (Nov 2023) API access | OpenAI | 15.9% ± 0.7% | – |
| 93 | ministral-8b-2410 Open weights (non-commercial) | Mistral AI | 14.9% ± 0.7% | – |
| 94 | claude-3-haiku-20240307 API access | Anthropic | 14.9% ± 0.7% | – |
| 95 | ministral-3b-2410 API access | Mistral AI | 14.4% ± 0.7% | – |
| 96 | claude-2.0 API access | Anthropic | 11.7% ± 0.7% | – |
| 97 | dbrx-instruct Open weights (restricted use) | Databricks | 11.7% ± 0.7% | – |
| 98 | GPT-3.5 Turbo (Jan 2024) API access | OpenAI | 11.6% ± 0.6% | – |
| 99 | gemini-1.0-pro-001 API access | Google DeepMind | 11.2% ± 0.6% | – |
| 100 | open-mistral-nemo-2407 Open weights (unrestricted) | Mistral AI | 10.8% ± 0.6% | – |
