MATH level 5 leaderboard
- GPT-5 leads
- 98.1%
- models measured
- 28
- lead over #2
- +0.3 pts
- median
- 72.0%
Leader on MATH level 5 · read October 22
On October 22, 2026, 44 of our catalog's models had a MATH level 5 result, and OpenAI GPT-5 led with 98.1%, ahead of OpenAI GPT-5 Mini at 97.8%.
The hardest level of the MATH competition dataset. Share solved.
Role in the index
MATH level 5 is shown as published and does not feed any of our indexes. Read the methodology
| # | Model | Result | Evidence | Coding index | Output $/Mtok |
|---|---|---|---|---|---|
| 1 | GPT-5OpenAI | 98.1% | 38.8 | $10.00 | |
| 2 | GPT-5 MiniOpenAI | 97.8% | 34.2 | $2.00 | |
| 3 | o4 MiniOpenAI | 97.8% | Not benchmarked | $4.40 | |
| 4 | o3OpenAI | 97.8% | Not benchmarked | $8.00 | |
| 5 | Claude Sonnet 4.5Anthropic | 97.7% | 39.5 | $15.00 | |
| 6 | R1 0528DeepSeek | 96.6% | Not benchmarked | $2.18 | |
| 7 | o3 MiniOpenAI | 96.5% | Not benchmarked | $4.40 | |
| 8 | Claude Haiku 4.5Anthropic | 96.4% | 35.6 | $5.00 | |
| 9 | GPT-5 NanoOpenAI | 95.2% | Not benchmarked | $0.40 | |
| 10 | o1OpenAI | 94.7% | Not benchmarked | $60.00 | |
| 11 | R1RetiredDeepSeek | 93.1% | Not benchmarked | $2.50 | |
| 12 | Claude 3.7 SonnetRetiredAnthropic | 91.2% | Not benchmarked | $15.00 | |
| 13 | o1-miniRetiredOpenAI | 89.2% | Not benchmarked | $4.40 | |
| 14 | Grok 3RetiredxAI | 88.7% | Not benchmarked | $15.00 | |
| 15 | GPT-4.1 MiniOpenAI | 87.3% | Not benchmarked | $1.60 | |
| 16 | Claude Opus 4RetiredAnthropic | 85.0% | Not benchmarked | $75.00 | |
| 17 | GPT-4.1OpenAI | 83.0% | Not benchmarked | $8.00 | |
| 18 | Gemini 2.0 FlashRetiredGoogle | 82.2% | Not benchmarked | $0.40 | |
| 19 | DeepSeek V3 0324DeepSeek | 75.5% | Not benchmarked | $1.07 | |
| 20 | Gemma 3 27BGoogle | 74.0% | Not benchmarked | $0.30 | |
| 21 | Gemini 1.5 ProRetiredGoogle | 70.4% | Not benchmarked | $5.00 | |
| 22 | GPT-4.1 NanoOpenAI | 70.0% | Not benchmarked | $0.40 | |
| 23 | Qwen-PlusQwen | 65.3% | Not benchmarked | $0.78 | |
| 24 | Phi 4Microsoft | 64.9% | Not benchmarked | $0.14 | |
| 25 | Grok 2RetiredxAI | 63.5% | Not benchmarked | $10.00 | |
| 26 | Qwen2.5 72B InstructQwen | 63.2% | Not benchmarked | $0.40 | |
| 27 | Llama 4 ScoutMeta Llama | 62.3% | Not benchmarked | $0.50 | |
| 28 | Gemini 1.5 FlashRetiredGoogle | 61.9% | Not benchmarked | $0.30 | |
| 29 | Claude 3.5 SonnetRetiredAnthropic | 56.9% | Not benchmarked | $15.00 | |
| 30 | GPT-4o (2024-08-06)OpenAI | 53.3% | Not benchmarked | $10.00 | |
| 31 | GPT-4o-mini (2024-07-18)OpenAI | 52.6% | Not benchmarked | $0.60 | |
| 32 | GPT-4o (2024-05-13)OpenAI | 51.0% | Not benchmarked | $15.00 | |
| 33 | GPT-4o (2024-11-20)OpenAI | 49.8% | Not benchmarked | $10.00 | |
| 34 | Claude 3.5 HaikuRetiredAnthropic | 46.4% | Not benchmarked | $4.00 | |
| 35 | Mistral Large 2407Mistral AI | 44.8% | Not benchmarked | $6.00 | |
| 36 | Llama 3.3 70B InstructMeta Llama | 41.6% | Not benchmarked | $0.72 | |
| 37 | Claude 3 OpusRetiredAnthropic | 37.5% | Not benchmarked | $75.00 | |
| 38 | Llama 3.1 70B InstructMeta Llama | 36.7% | Not benchmarked | $0.56 | |
| 39 | Gemma 2 27BGoogle | 27.9% | Not benchmarked | $0.65 | |
| 40 | WizardLM-2 8x22BRetiredMicrosoft | 25.7% | Not benchmarked | $0.62 | |
| 41 | Llama 3.1 8B InstructMeta Llama | 22.9% | Not benchmarked | $0.08 | |
| 42 | Claude 3 SonnetRetiredAnthropic | 18.2% | Not benchmarked | $15.00 | |
| 43 | Claude 3 HaikuRetiredAnthropic | 14.9% | Not benchmarked | $1.25 | |
| 44 | Claude 2RetiredAnthropic | 11.7% | Not benchmarked | $24.00 |
Only models in our catalog are listed, and each appears once with its best published run. 215 rows were read from the source on October 22, 2026, and rows that match no catalog model are left out.