LiveCodeBench — Large pooled entry (82 models), no version qualifier; kept sep. from versioned per rule.
| # | Model | Vendor | Best score | Runs | Last seen |
|---|---|---|---|---|---|
| 1 | DeepSeek-V4-Pro | DeepSeek | 93.5 | 10 | 2026-08-23 |
| 2 | Gemini 3 Pro | 92 | 1 | 2026-06-05 | |
| 3 | Gemini 3.1 Pro | 91.7 | 4 | 2026-07-04 | |
| 4 | DeepSeek-V4-Flash | DeepSeek | 91.6 | 4 | 2026-06-27 |
| 5 | Kimi K2.6 | Moonshot | 89.6 | 4 | 2026-07-04 |
| 6 | GPT-5.2 | OpenAI | 89 | 1 | 2026-06-05 |
| 7 | Claude Opus 4.6 | Anthropic | 88.8 | 4 | 2026-07-04 |
| 8 | Solar Pro4 | Upstage | 87.8 | 1 | 2026-08-23 |
| 9 | Claude Opus 4.5 | Anthropic | 87 | 1 | 2026-06-05 |
| 10 | DeepSeek-V3.2 | DeepSeek | 86 | 3 | 2026-08-23 |
| 11 | GPT-5 | OpenAI | 85 | 3 | 2026-06-25 |
| 12 | o3 | OpenAI | 84.7 | 1 | 2026-05-03 |
| 13 | o4-mini | OpenAI | 84.5 | 3 | 2026-05-03 |
| 14 | DeepSeek-R1 | DeepSeek | 84.4 | 8 | 2026-08-23 |
| 15 | Gemini 2.5 Pro Preview 06.05 (32k think) | 84.3 | 1 | 2026-05-30 | |
| 16 | MiniMax M2 | MiniMax | 83 | 3 | 2026-08-23 |
| 17 | Gemini 2.5 Pro | 82.7 | 6 | 2026-06-06 | |
| 18 | Nemotron 3 Super 120B A12B | NVIDIA | 82.1 | 2 | 2026-08-23 |
| 19 | NVIDIA-Nemotron-Labs-3-Puzzle-75B-A9B-BF16 | NVIDIA | 81.1 | 1 | 2026-07-07 |
| 20 | MiniMax M2.1 | MiniMax | 81 | 2 | 2026-08-23 |
| 21 | OpenReasoning-Nemotron-32B | NVIDIA | 81 | 1 | 2026-05-03 |
| 22 | EXAONE 4.0 32B | LG AI | 80.9 | 1 | 2026-05-03 |
| 23 | Grok 3 Mini Reasoning | SpaceXAI | 80.4 | 3 | 2026-08-23 |
| 24 | Qwen3 235B A22B | Alibaba | 80.4 | 4 | 2026-08-23 |
| 25 | Grok 4 Fast | SpaceXAI | 80 | 1 | 2026-08-23 |
| 26 | Gemma 4 26B A4B | 79.8 | 1 | 2026-05-06 | |
| 27 | GLM-4.6 | Z.ai | 79.5 | 3 | 2026-06-25 |
| 28 | Grok 3 | SpaceXAI | 79.4 | 2 | 2026-08-23 |
| 29 | Kimi K2 (Reasoning) | Moonshot | 79.2 | 3 | 2026-06-25 |
| 30 | Grok 4 | SpaceXAI | 79 | 1 | 2026-08-23 |
| 31 | OpenCodeReasoning-Nemotron-1.1-32B | NVIDIA | 78.5 | 1 | 2026-05-03 |
| 32 | PaCoRe-8B | StepFun | 78.2 | 1 | 2026-06-25 |
| 33 | o3-mini | OpenAI | 77.7 | 5 | 2026-06-18 |
| 34 | Step3 VL 10B | StepFun | 76.4 | 2 | 2026-06-05 |
| 35 | Gemini 2.5 Flash | 76.2 | 5 | 2026-06-20 | |
| 36 | DeepSeek-V3.1-Terminus | DeepSeek | 74.9 | 2 | 2026-06-25 |
| 37 | DeepSeek-V3.1 | DeepSeek | 74.8 | 3 | 2026-08-23 |
| 38 | Qwen3 235B Thinking | Alibaba | 74.5 | 2 | 2026-06-25 |
| 39 | DeepSeek-V3.2-Exp | DeepSeek | 74.1 | 1 | 2026-08-23 |
| 40 | GLM-4.5 | Z.ai | 72.9 | 1 | 2026-08-23 |
| 41 | NVIDIA Nemotron Nano 9B V2 | NVIDIA | 71.1 | 1 | 2026-08-23 |
| 42 | Claude Sonnet 4.5 | Anthropic | 71 | 2 | 2026-06-06 |
| 43 | GLM-4.5-Air | Z.ai | 70.7 | 1 | 2026-08-23 |
| 44 | RLVR-8B | StepFun | 70.6 | 2 | 2026-06-25 |
| 45 | Claude 4 Opus | Anthropic | 70.4 | 2 | 2026-05-29 |
| 46 | Qwen3 VL 235B A22B Reasoning | Alibaba | 69.5 | 2 | 2026-06-12 |
| 47 | Claude Sonnet 4 | Anthropic | 68.5 | 3 | 2026-06-06 |
| 48 | Mercury 2 | Inception | 67 | 1 | 2026-08-23 |
| 49 | Llama 3.1 Nemotron Ultra 253B V1 | NVIDIA | 66.3 | 1 | 2026-08-23 |
| 50 | Qwen3 32B | Alibaba | 65.7 | 1 | 2026-08-23 |
| 51 | MiniMax M1 80K | MiniMax | 65 | 1 | 2026-08-23 |
| 52 | Ministral 3 14B | Mistral | 64.6 | 4 | 2026-06-12 |
| 53 | Mistral Small 4 | Mistral | 63.6 | 1 | 2026-08-23 |
| 54 | Qwen3 30B A3B | Alibaba | 62.6 | 1 | 2026-08-23 |
| 55 | MiniMax M1 40K | MiniMax | 62.3 | 1 | 2026-08-23 |
| 56 | Ministral 3 8B | Mistral | 61.6 | 4 | 2026-06-12 |
| 57 | DeepSeek-R1 0528-Qwen3-8B | DeepSeek | 60.5 | 2 | 2026-06-04 |
| 58 | Qwen3 14B | Alibaba | 59.3 | 4 | 2026-06-05 |
| 59 | Qwen3 VL Thinking (8B) | Alibaba | 58 | 4 | 2026-06-05 |
| 60 | DeepSeek-R1-Distill-Llama-70B | DeepSeek | 57.5 | 1 | 2026-08-23 |
| 61 | DeepSeek-R1-Distill-Qwen-32B | DeepSeek | 57.2 | 2 | 2026-08-23 |
| 62 | Grok 3 Beta | SpaceXAI | 57 | 1 | 2026-07-13 |
| 63 | Qwen2.5 Instruct 72B | Alibaba | 55.5 | 3 | 2026-08-23 |
| 64 | Ministral 3 3B | Mistral | 54.8 | 4 | 2026-06-12 |
| 65 | Kimi K2 Instruct | Moonshot | 53.7 | 1 | 2026-08-23 |
| 66 | DeepSeek-R1-Distill-Qwen-14B | DeepSeek | 53.1 | 1 | 2026-08-23 |
| 67 | Phi 4 Reasoning Plus | Microsoft | 53.1 | 1 | 2026-08-23 |
| 68 | Qwen3 VL 4B (Reasoning) | Alibaba | 51.3 | 4 | 2026-06-05 |
| 69 | Magistral Medium 1 | Mistral | 50.3 | 1 | 2026-08-23 |
| 70 | R1-Zero | DeepSeek | 50 | 1 | 2026-08-23 |
| 71 | QwQ 32B Preview | Alibaba | 50 | 1 | 2026-08-23 |
| 72 | DeepSeek-V3 | DeepSeek | 49.6 | 6 | 2026-08-23 |
| 73 | GLM-4.6V (106B-A12B) | Z.ai | 48.7 | 2 | 2026-06-05 |
| 74 | Llama 4 Maverick | Meta | 43.4 | 2 | 2026-08-23 |
| 75 | DeepSeek-V2.5 | DeepSeek | 41.8 | 2 | 2026-08-23 |
| 76 | Grok 3 Mini Beta | SpaceXAI | 41.5 | 1 | 2026-07-13 |
| 77 | SynLogic Mix 3 32B | MiniMax | 40.7 | 1 | 2026-06-05 |
| 78 | DeepSeek-R1-Zero-Qwen-32B | DeepSeek | 40.2 | 1 | 2026-06-05 |
| 79 | DeepSeek-R1-Distill-Llama-8B | DeepSeek | 39.6 | 1 | 2026-08-23 |
| 80 | GPT-4o | OpenAI | 38.3 | 2 | 2026-05-03 |
| 81 | DeepSeek-R1-Distill-Qwen-7B | DeepSeek | 37.6 | 1 | 2026-08-23 |
| 82 | GPT-4 Turbo | OpenAI | 37.3 | 1 | 2026-05-03 |
| 83 | GPT-4o mini | OpenAI | 35.5 | 1 | 2026-05-03 |
| 84 | Gemini 2.0 Flash (Reasoning) | 35.1 | 1 | 2026-08-23 | |
| 85 | Mistral Large 3 675B Base 2512 | Mistral | 34.4 | 1 | 2026-08-23 |
| 86 | Mistral Large | Mistral | 34.4 | 1 | 2026-08-23 |
| 87 | Gemini 2.5 Flash Lite | 33.7 | 1 | 2026-08-23 | |
| 88 | Llama 3.3 70B Instruct | Meta | 33.3 | 1 | 2026-05-01 |
| 89 | Claude 3.5 Sonnet | Anthropic | 32.8 | 1 | 2026-05-03 |
| 90 | Llama 4 Scout | Meta | 32.8 | 3 | 2026-08-23 |
| 91 | Qwen 2.5 Coder 32B Instruct | Alibaba | 31.4 | 1 | 2026-08-23 |
| 92 | Gemini 1.5 Pro | 30.5 | 1 | 2026-05-03 | |
| 93 | Llama 3.1 Instruct 405B | Meta | 30.1 | 3 | 2026-05-03 |
| 94 | Gemma 3 27B | 29.7 | 1 | 2026-08-23 | |
| 95 | Qwen2 7B Instruct | Alibaba | 26.6 | 1 | 2026-08-23 |
| 96 | Gemma 3 12B | 24.6 | 1 | 2026-08-23 | |
| 97 | Yi Coder 9B Chat | 01.AI | 23 | 3 | 2026-07-12 |
| 98 | Claude 3 Haiku | Anthropic | 22.5 | 1 | 2026-05-03 |
| 99 | DeepSeekCoder-33B-Ins | DeepSeek | 22.3 | 3 | 2026-07-12 |
| 100 | DeepSeek-V2 | DeepSeek | 20.3 | 2 | 2026-05-03 |
Best tracked score per model (default configuration; source-attributed and verification-tiered). Open a model for its full benchmark surface, provenance and pricing.