GSM8K (Grade School Math 8K) — grade-school math word problems; the standard accuracy benchmark.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | MiMo V2.5 Pro | Xiaomi | 99.6 | 2 | 2026-08-23 |
| 2 | MiMo V2.5 Pro Base | Xiaomi | 99.6 | 3 | 2026-06-05 |
| 3 | Kimi K2 Instruct | Moonshot | 97.3 | 1 | 2026-08-23 |
| 4 | o1 | OpenAI | 97.1 | 1 | 2026-08-23 |
| 5 | GPT-4.5 Preview | OpenAI | 97 | 1 | 2026-08-23 |
| 6 | Claude 3.5 Sonnet | Anthropic | 96.9 | 3 | 2026-08-23 |
| 7 | Llama 3.1 Instruct 405B | Meta | 96.8 | 5 | 2026-08-23 |
| 8 | DeepSeek-V3 | DeepSeek | 96.7 | 2 | 2026-07-02 |
| 9 | Gemma 3 27B | 95.9 | 1 | 2026-08-23 | |
| 10 | Qwen2.5 Instruct 72B | Alibaba | 95.8 | 3 | 2026-08-23 |
| 11 | GPT-4o | OpenAI | 95.6 | 3 | 2026-07-02 |
| 12 | Gemini 2.0 Flash Exp | 95.4 | 2 | 2026-06-06 | |
| 13 | Hy3 preview-Base | Tencent | 95.4 | 2 | 2026-06-05 |
| 14 | Gemini 1.5 Pro | 95.2 | 4 | 2026-08-23 | |
| 15 | DeepSeek-V2.5 | DeepSeek | 95.1 | 1 | 2026-08-23 |
| 16 | Claude 3 Opus | Anthropic | 95 | 2 | 2026-08-23 |
| 17 | Nova Pro (Non-Reasoning) | Amazon | 94.8 | 3 | 2026-08-23 |
| 18 | MiniMax Text 01 | MiniMax | 94.8 | 1 | 2026-06-06 |
| 19 | Gemini 2.0 Flash (experimental) | 94.6 | 1 | 2026-07-02 | |
| 20 | Nova Lite (Non-Reasoning) | Amazon | 94.5 | 1 | 2026-08-23 |
| 21 | Qwen3.5 122B A10B | Alibaba | 94.5 | 1 | 2026-07-05 |
| 22 | Gemma 3 12B | 94.4 | 1 | 2026-08-23 | |
| 23 | Qwen3 235B A22B | Alibaba | 94.4 | 1 | 2026-08-23 |
| 24 | Llama 3.3 Instruct Turbo (70B) | Meta | 94.2 | 1 | 2026-07-02 |
| 25 | Granite 4.1 30B | IBM | 94.2 | 1 | 2026-06-15 |
| 26 | Llama 3.1 Instruct Turbo (70B) | Meta | 93.8 | 1 | 2026-07-02 |
| 27 | Llama 3.2 Vision Instruct Turbo (90B) | Meta | 93.6 | 1 | 2026-07-02 |
| 28 | Kimi K2 Base | Moonshot | 93.5 | 11 | 2026-08-23 |
| 29 | GPT4 | OpenAI | 93.2 | 3 | 2026-08-12 |
| 30 | Mistral Large 2 | Mistral | 93 | 2 | 2026-08-23 |
| 31 | Hunyuan Large | Tencent | 92.8 | 1 | 2026-05-31 |
| 32 | DeepSeek-V4-Pro-Base | DeepSeek | 92.6 | 3 | 2026-06-05 |
| 33 | Granite 4.1 8B | IBM | 92.5 | 1 | 2026-06-15 |
| 34 | Nova Micro (Non-Reasoning) | Amazon | 92.3 | 3 | 2026-08-23 |
| 35 | Claude 3 Sonnet | Anthropic | 92.3 | 2 | 2026-08-23 |
| 36 | MiMo V2 Flash Base | Xiaomi | 92.3 | 4 | 2026-06-13 |
| 37 | Qwen2 Instruct (72B) | Alibaba | 92 | 2 | 2026-08-23 |
| 38 | NVIDIA Nemotron 3 Nano 30B A3B | NVIDIA | 91.4 | 1 | 2026-08-11 |
| 39 | Llama 3.1 Nemotron Instruct 70B | NVIDIA | 91.4 | 1 | 2026-08-23 |
| 40 | DeepSeek-V3.1-Base | DeepSeek | 91.4 | 4 | 2026-06-13 |
| 41 | Nemotron 3.5 Lightning | NVIDIA | 91.3 | 1 | 2026-08-11 |
| 42 | NVIDIA-Nemotron-3.5-Lightning-30B-A3B-Base-BF16 | NVIDIA | 91.3 | 1 | 2026-08-24 |
| 43 | Mistral Large 3 675B Base 2512 | Mistral | 91.2 | 1 | 2026-06-05 |
| 44 | DeepSeek-V3.2-Exp-Base | DeepSeek | 91.1 | 5 | 2026-06-13 |
| 45 | Qwen 2.5 Coder 32B Instruct | Alibaba | 91.1 | 1 | 2026-08-23 |
| 46 | DeepSeek-V4-Flash-Base | DeepSeek | 90.8 | 3 | 2026-06-05 |
| 47 | Nemotron 3 Super 120B A12B | NVIDIA | 90.7 | 1 | 2026-08-11 |
| 48 | Palmyra-X-004 | Writer | 90.5 | 1 | 2026-07-02 |
| 49 | Qwen3.5 35B A3B | Alibaba | 90.1 | 1 | 2026-08-11 |
| 50 | GLM-4.5-Base | Z.ai | 90.1 | 5 | 2026-06-05 |
| 51 | Gemma 3 4B | 89.9 | 2 | 2026-08-23 | |
| 52 | Claude 3 Haiku | Anthropic | 88.9 | 2 | 2026-08-23 |
| 53 | Phi 3.5 MoE Instruct | Microsoft | 88.7 | 1 | 2026-08-23 |
| 54 | Qwen2.5 Omni 7B | Alibaba | 88.7 | 1 | 2026-08-23 |
| 55 | Phi 4 Mini Instruct | Microsoft | 88.6 | 1 | 2026-08-23 |
| 56 | Step3.5 Flash Base | StepFun | 88.2 | 2 | 2026-06-05 |
| 57 | DeepSeek-V3-Base | DeepSeek | 88.2 | 2 | 2026-06-05 |
| 58 | NVIDIA-Nemotron-3-Ultra-550B-A55B-Base-BF16 | NVIDIA | 88.1 | 1 | 2026-08-24 |
| 59 | Nemotron 3 Ultra 550B A55B Base | NVIDIA | 88.1 | 1 | 2026-06-05 |
| 60 | Claude 2 | Anthropic | 88 | 3 | 2026-08-12 |
| 61 | Phi 3 Medium 4K Instruct | Microsoft | 87.8 | 2 | 2026-07-03 |
| 62 | Solar Pro | Upstage | 87.1 | 1 | 2026-07-02 |
| 63 | Jamba 1.5 Large | AI21 | 87 | 2 | 2026-08-23 |
| 64 | Phi 3.5 Mini Instruct | Microsoft | 86.2 | 1 | 2026-08-23 |
| 65 | Qwen3 1.7B | Alibaba | 85.6 | 2 | 2026-08-09 |
| 66 | lfm-2.5-1.2b-thinking:free | Liquid AI | 85.6 | 1 | 2026-08-09 |
| 67 | Claude 1.3 | Anthropic | 85.2 | 1 | 2026-07-17 |
| 68 | GPT-4o mini | OpenAI | 84.3 | 1 | 2026-07-02 |
| 69 | MAI-Thinking-1 | Microsoft | 84 | 1 | 2026-08-14 |
| 70 | Granite 4.1 30B Base | IBM | 83.8 | 2 | 2026-06-15 |
| 71 | Llama 3.1 70B Instruct | Meta | 83.7 | 1 | 2026-05-31 |
| 72 | MiMo V2.5 Base | Xiaomi | 83.3 | 3 | 2026-06-05 |
| 73 | Llama 3.1 8B Instruct | Meta | 83.2 | 1 | 2026-06-11 |
| 74 | Palmyra X V3 (72B) | Writer | 83.1 | 1 | 2026-07-02 |
| 75 | Qwen2.5 Instruct Turbo (7B) | Alibaba | 83 | 1 | 2026-07-02 |
| 76 | Amazon Nova Lite | Amazon | 82.9 | 1 | 2026-07-02 |
| 77 | LFM2 2.6B | Liquid AI | 82.4 | 1 | 2026-08-09 |
| 78 | GPT-4 Turbo | OpenAI | 82.4 | 2 | 2026-07-02 |
| 79 | Qwen2 7B Instruct | Alibaba | 82.3 | 1 | 2026-08-23 |
| 80 | Llama 3.2 Vision Instruct Turbo (11B) | Meta | 82.3 | 1 | 2026-07-02 |
| 81 | Granite 3.2 8B Instruct | IBM | 81.7 | 1 | 2026-06-11 |
| 82 | Claude 3.5 Haiku | Anthropic | 81.5 | 2 | 2026-07-02 |
| 83 | Gemma 2 Instruct (27B) | 81.2 | 1 | 2026-07-02 | |
| 84 | Granite 3.3 8B Instruct | IBM | 80.9 | 2 | 2026-08-23 |
| 85 | Llama 3 (70B) | Meta | 80.5 | 1 | 2026-07-02 |
| 86 | Llama 3.1 Instruct Turbo (8B) | Meta | 79.8 | 1 | 2026-07-02 |
| 87 | DeepSeek-Chat | DeepSeek | 79.5 | 1 | 2026-07-02 |
| 88 | DeepSeek-V2 | DeepSeek | 79.2 | 1 | 2026-05-31 |
| 89 | Granite 3.1 8B Instruct | IBM | 79.2 | 1 | 2026-06-11 |
| 90 | Qwen2.5 3B | Alibaba | 79.1 | 2 | 2026-06-05 |
| 91 | Gemini 1.5 Flash (001) | 78.5 | 2 | 2026-07-02 | |
| 92 | DeepSeek-R1-Distill-Qwen-7B | DeepSeek | 78.5 | 1 | 2026-06-11 |
| 93 | Mistral Nemo | Mistral | 78.2 | 1 | 2026-07-02 |
| 94 | Llama 3.2 3B Instruct | Meta | 77.7 | 4 | 2026-08-23 |
| 95 | Moonlight | Moonshot | 77.4 | 2 | 2026-06-12 |
| 96 | Gemma 4 26B A4B | 77 | 1 | 2026-08-11 | |
| 97 | Arctic Instruct | Snowflake | 76.8 | 1 | 2026-07-02 |
| 98 | Gemma 2 Instruct (9B) | 76.2 | 2 | 2026-08-23 | |
| 99 | Jamba 1.5 Mini | AI21 | 75.8 | 2 | 2026-08-23 |
| 100 | Gemma 2 27B IT | 74 | 1 | 2026-08-23 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.