Reasoning
LiveBench Reasoning Leaderboard
LiveBench Reasoning measures logical, multi-step reasoning using contamination-free questions that are refreshed regularly, so models cannot have trained on the test set.
Source: livebench19 open models ranked+51 proprietaryData through Jun 2026
All models ranked on LiveBench Reasoning
Proprietary / closed models are shown dimmed — you can't run them locally, but they show where the open field stands.
| # | Model | Score | VRAM @ Q4 |
|---|---|---|---|
| 1 | GPT 6 Astra Max · proprietary | 92.7 | API only |
| 2 | GPT 6.1 Sol Max · proprietary | 92.6 | API only |
| 3 | Claude Opus 5.5 Max · proprietary | 92.2 | API only |
| 4 | Claude Fable 5.1 Max · proprietary | 91.7 | API only |
| 5 | GPT 5.6 Sol Max · proprietary | 91.7 | API only |
| 6 | Claude Sonnet 5.5 Max · proprietary | 91.6 | API only |
| 7 | GPT 6.1 Sol · proprietary | 91.6 | API only |
| 8 | Claude Opus 5 Max · proprietary | 91.2 | API only |
| 9 | Kimi K3 · 2779.9B | 90.7 | ~1674 GB |
| 10 | Claude Opus 5.5 · proprietary | 90.7 | API only |
| 11 | GPT 5.6 Terra Max · proprietary | 90.6 | API only |
| 12 | Muse Wonder 1.0 Max · proprietary | 90.6 | API only |
| 13 | Grok 4.6 · proprietary | 90.5 | API only |
| 14 | Smaug Agentic · proprietary | 90.3 | API only |
| 15 | Muse Spark 1.2 · proprietary | 90.0 | API only |
| 16 | Claude Fable 5 Max · proprietary | 89.7 | API only |
| 17 | GPT 5.5 · proprietary | 89.7 | API only |
| 18 | Muse Spark 1.3 · proprietary | 89.7 | API only |
| 19 | Gemini 3.8 Flash · proprietary | 89.3 | API only |
| 20 | Claude Opus 4.8 Max · proprietary | 89.2 | API only |
| 21 | Claude Sonnet 5 · proprietary | 88.7 | API only |
| 22 | Claude Opus 4.6 · proprietary | 88.7 | API only |
| 23 | GPT 6 Sol Max · proprietary | 88.7 | API only |
| 24 | Qwen3.8 Max · proprietary | 88.2 | API only |
| 25 | GPT 5.4 · proprietary | 88.1 | API only |
| 26 | Gemini 3.7 Flash · proprietary | 87.8 | API only |
| 27 | Muse Spark 1.1 · proprietary | 87.7 | API only |
| 28 | Qwen3.8 Flash Next · 180.0B | 87.4 | ~108 GB |
| 29 | Claude Opus 4.7 · proprietary | 87.2 | API only |
| 30 | Grok 4.5 · proprietary | 87.2 | API only |
| 31 | Claude Sonnet 5.5 · proprietary | 86.8 | API only |
| 32 | DeepSeek V4.1 Flash · 763.2B | 86.7 | ~458 GB |
| 33 | DeepSeek V4 Flash 0731 · 304.2B | 86.6 | ~183 GB |
| 34 | Smaug Flash · 304.2B | 86.2 | ~609 GB |
| 35 | DeepSeek V4 Pro 0813 · 1650.5B | 85.8 | ~991 GB |
| 36 | GLM 5.3 · 753.3B | 85.8 | ~456 GB |
| 37 | GPT 5.6 Luna Max · proprietary | 85.6 | API only |
| 38 | DeepSeek V4 Flash Vision Exp · 304.6B | 85.4 | ~183 GB |
| 39 | Gemini 3.6 Flash · proprietary | 85.2 | API only |
| 40 | Claude Sonnet 4.6 · proprietary | 84.8 | API only |
| 41 | Claude Haiku 5.5 Max · proprietary | 84.6 | API only |
| 42 | Gemini 3.1 Pro · proprietary | 84.0 | API only |
| 43 | Mistral Large 4 · proprietary | 83.9 | API only |
| 44 | Qwen3.7 Max · proprietary | 83.3 | API only |
| 45 | GPT 5.2 · proprietary | 83.2 | API only |
| 46 | Kimi K2.7 Code · 1026.9B | 82.8 | ~620 GB |
| 47 | Smaug Mini · proprietary | 82.8 | API only |
| 48 | DeepSeek V4 Pro · 1598.8B | 82.7 | ~960 GB |
| 49 | Grok 4.7 · proprietary | 82.7 | API only |
| 50 | Gemini 3.5 Flash · proprietary | 82.0 | API only |
| 51 | GPT 6 Luna Max · proprietary | 81.8 | API only |
| 52 | Claude Haiku 5.5 · proprietary | 81.2 | API only |
| 53 | GPT 5.4 Nano · proprietary | 81.1 | API only |
| 54 | Claude Opus 4.5 · proprietary | 80.1 | API only |
| 55 | Qwen3.8 27B · 27.8B | 80.0 | ~17 GB |
| 56 | Kimi K2.6 · 1026.9B | 79.4 | ~620 GB |
| 57 | GLM 5.2 · 753.3B | 78.6 | ~456 GB |
| 58 | Inkling · 952.4B | 78.3 | ~572 GB |
| 59 | GPT 5.2 Codex · proprietary | 77.7 | API only |
| 60 | GLM 5.3 Flash · 321.3B | 77.6 | ~195 GB |
| 61 | Ox Alpha Max · proprietary | 76.6 | API only |
| 62 | Grok Build 0.1 · proprietary | 76.4 | API only |
| 63 | Qwen3.6 Plus · proprietary | 75.8 | API only |
| 64 | NVIDIA Nemotron 3 Ultra 550B A55B BF16 · 560.5B | 74.7 | ~370 GB |
| 65 | MiniMax M3 · 427.0B | 74.5 | ~257 GB |
| 66 | GPT 5.4 Mini · proprietary | 71.3 | API only |
| 67 | Grok 4.3 · proprietary | 70.8 | API only |
| 68 | DeepSeek V4 Flash · 290.9B | 70.6 | ~175 GB |
| 69 | Qwen3.6 27B · 27.8B | 70.3 | ~17 GB |
| 70 | Gemini 3.5 Flash Lite · proprietary | 60.2 | API only |
Score vs model size
Which models give the most quality for their size — the ones worth running locally.
- Qwen3.8 27B, 28B, score 80.0 — on the efficiency frontier (best score at its size or smaller).
- Qwen3.8 Flash Next, 180B, score 87.4 — on the efficiency frontier (best score at its size or smaller).
- Kimi K3, 2.8T, score 90.7 — on the efficiency frontier (best score at its size or smaller).
LiveBench Reasoning: frequently asked questions
- What is the best open LLM on LiveBench Reasoning?
- Kimi K3 is the top open model on LiveBench Reasoning, scoring 90.7. Among all models tested — including proprietary ones — it ranks #9. The top model overall is GPT 6 Astra Max (OpenAI) at 92.7.
- What's the best LiveBench Reasoning model you can run on a 24 GB GPU?
- Qwen3.8 27B is the highest-scoring open model that fits in 24 GB at 4-bit quantization (~17 GB), scoring 80.0 on LiveBench Reasoning.
- Can open models match proprietary models on LiveBench Reasoning?
- Not quite on LiveBench Reasoning: the strongest proprietary model (GPT 6 Astra Max) scores 92.7, ahead of the best open model (Kimi K3) at 90.7 — but you can run the open one yourself.
Scores aggregated from livebench. llmrun does not run this benchmark — see the source for methodology, or the about benchmarks for what it measures.