Best LLMs for math: open and proprietary
Open and proprietary models together on one 0–100 scale (mathematics). Switch to local models to see only what you can download and run.
40 benchmarks4 sourcesUpdated 3 Oct 2026Reference scale 2026-Q4How the score is built
| # | Model | llmrun Score | Coding | Agents | Math | Science | Reasoning | Benchmarks | VRAM |
|---|---|---|---|---|---|---|---|---|---|
| 121 | Gemini 1.5 Pro 0015 benchmarks | – | – | 19 | – | 20 | 5 | – | |
| 122 | Llama 3.2 90B Vision Instructopen58.5 GB · 4 benchmarks | – | – | 18 | – | – | 4 | 58.5 GB | |
| 123 | Claude 3 Opus (Feb 29, 2024)10 benchmarks | – | – | 18 | – | 22 | 10 | – | |
| 124 | Llama 3.1 70B Instructopen46.6 GB · 8 benchmarks | – | – | 18 | – | 22 | 8 | 46.6 GB | |
| 125 | Gemma 2 27B ITopen18.0 GB · 4 benchmarks | – | – | 16 | – | – | 4 | 18.0 GB | |
| 126 | Gemini 1.5 Flash 0015 benchmarks | – | – | 15 | – | 16 | 5 | – | |
| 127 | Mistral Large 24024 benchmarks | – | – | 15 | – | – | 4 | – | |
| 128 | Meta Llama 3 70B Instructopen46.6 GB · 5 benchmarks | – | – | 14 | – | – | 5 | 46.6 GB | |
| 129 | GPT 4 (Jun 13)9 benchmarks | – | – | 14 | – | 23 | 9 | – | |
| 130 | Hermes 2 Theta Llama 3 70Bopen43.3 GB · 3 benchmarks | — | – | – | 14 | – | – | 3 | 43.3 GB |
| 131 | Llama 3.1 8B Instructopen5.3 GB · 11 benchmarks | 14 | – | 14 | 7 | 16 | 11 | 5.3 GB | |
| 132 | Gemma 2 9B ITopen6.1 GB · 4 benchmarks | – | – | 14 | – | – | 4 | 6.1 GB | |
| 133 | Claude 3 Sonnet (Feb 29, 2024)5 benchmarks | – | – | 13 | – | – | 5 | – | |
| 134 | Claude 3 Haiku (Mar 07, 2024)7 benchmarks | – | – | 12 | – | 15 | 7 | – | |
| 135 | Claude 2.04 benchmarks | – | – | 12 | – | – | 4 | – | |
| 136 | GPT 3.5 Turbo (Jan 25)10 benchmarks | – | – | 12 | – | 12 | 10 | – | |
| 137 | Gemini 1.0 Pro 0014 benchmarks | – | – | 11 | – | – | 4 | – | |
| 138 | Deepseek Llm 67B Chatopen41.3 GB · 4 benchmarks | – | – | 10 | – | – | 4 | 41.3 GB | |
| 139 | Meta Llama 3 8B Instructopen5.3 GB · 7 benchmarks | – | – | 10 | – | 14 | 7 | 5.3 GB | |
| 140 | Mistral 7B Instruct v0.3open4.9 GB · 4 benchmarks | – | – | 10 | – | – | 4 | 4.9 GB | |
| 141 | Llama 2 70B Chat HFopen45.5 GB · 5 benchmarks | – | – | 9 | – | 10 | 5 | 45.5 GB |
The range after each score is a 90% interval: given the boards a model has results on, its true score is very likely inside it. Short ranges mean many agreeing results; long ranges mean few or conflicting ones.
Why some models are missing. A model gets a score only with results on at least 4 benchmarks across at least 2 categories, one of them reasoning or coding. Skill scores need at least 2 benchmarks in that skill; otherwise they show as –.
VRAM is llmrun's estimate at Q4_K_M (or the smallest quantization we track) for the weights plus a working context. Proprietary models can't run locally, so the VRAM filter hides them.
How the llmrun Score is built · llmrun does not run these benchmarks; scores are aggregated from public sources.