LiveBench 2.0 Reasoning
General QA · 2026-10-14
LiveBench 2.0 Reasoning suite: Single-answer reasoning questions (theory of mind, zebra puzzles, spatial, logic with navigation); 202 questions. LiveBench 2.0 (Abacus AI, released October 14, 2026) replaced the v1 question-answer categories with seven suites, five of them agentic; each suite is one run per model at its highest supported effort, a task the run could not measure scores 0, and a model without an input a suite needs (a text-only model on Browser) scores 0 there. Not comparable with LiveBench v1.
Top models (higher is better)
| Model | Score |
|---|---|
| Gemini 4 Argon | 95.1 |
| GPT-6 Astra | 92.7 |
| GPT-6.1 Sol | 92.6 |
| Claude Opus 5.5 | 92.2 |
| Claude Fable 5.1 | 91.7 |
| GPT-5.6 Sol | 91.7 |
| Claude Sonnet 5.5 | 91.6 |
| Claude Opus 5 | 91.2 |
| Grok 4.6 | 91.2 |
| Kimi K3.1 Pro | 91.1 |
| Kimi K3 | 90.7 |
| GPT-5.6 Terra | 90.6 |
| Muse Wonder | 90.6 |
| Muse Spark 1.2 | 90.0 |
| Claude Fable 5 | 89.7 |
| GPT-5.5 | 89.7 |
| Muse Spark 1.3 | 89.7 |
| Gemini 3.8 Flash | 89.3 |
| Opus 4.8 | 89.2 |
| DeepSeek V4.1 Flash 1017 | 89.1 |
| Sonnet 5 | 88.7 |
| GPT-6 Sol | 88.7 |
| Gemini 3.7 Flash | 87.8 |
| Qwen3.8-Flash-Next | 87.4 |
| Grok 4.5 | 87.2 |
| DeepSeek V4.1 Flash | 86.7 |
| Sonnet 4.6 | 86.4 |
| GLM-5.3 | 85.8 |
| GPT-5.6 Luna | 85.6 |
| DeepSeek V4 Flash Vision Exp | 85.4 |
| Gemini 3.1 Pro Preview | 84.0 |
| Mistral Large 4 Preview | 83.9 |
| Step 5 Preview | 83.8 |
| MiniMax M3.1 Flash | 82.9 |
| Grok 4.7 | 82.7 |
| Claude Haiku 5.5 | 82.4 |
| GPT-6 Luna | 81.8 |
| Beam 1.0 501B A23B | 80.2 |
| Qwen3.8 27B | 80.0 |
| MiMo-V2.6-Pro | 79.6 |