LiveBench
Contamination-free objective benchmark, refreshed monthly. Published by LiveBench (Apache-2.0) and redistributed by Epoch AI under CC-BY 4.0 in live_bench_external.csv (column 'Global average'). The Known Good aggregates published results; it runs no evaluations.
49 models scored
Higher is better
Published by LiveBench
Apache-2.0
LiveBench
LiveBench as measured and published by LiveBench (Apache-2.0). Ingested, normalised only where it feeds an index, and never re-run by us. Higher is better.
💡 A lightbulb marks a model its provider publishes as a reasoning model
The Known Good
Every model we hold a LiveBench score for
All 49 of 49 models
| # | Model | Creator | LiveBench | Known Good Index | Measured |
|---|---|---|---|---|---|
| 1 |
gemini-2.5-pro-exp-03-25
|
82.3 | — | 8 Oct 2026 | |
| 2 |
OpenAI: GPT-5.1
💡 |
OpenAI | 78.8 | 73 | 8 Oct 2026 |
| 3 |
claude-3-7-sonnet-20250219
|
Anthropic | 76.1 | 74 | 8 Oct 2026 |
| 4 |
OpenAI: o3 Mini
💡 |
OpenAI | 75.9 | 86 | 8 Oct 2026 |
| 5 |
OpenAI: o1
💡 |
OpenAI | 75.7 | 69 | 8 Oct 2026 |
| 6 |
QwQ-32B
|
Alibaba | 72.0 | 69 | 8 Oct 2026 |
| 7 |
DeepSeek: R1
💡 |
DeepSeek | 71.6 | 75 | 8 Oct 2026 |
| 8 |
gpt-4.5-preview
|
OpenAI | 69.0 | 54 | 8 Oct 2026 |
| 9 | GD gemini-2.0-flash-thinking-exp-01-21 | Google DeepMind,Google | 66.9 | 62 | 8 Oct 2026 |
| 10 |
DeepSeek: DeepSeek V3 0324
|
DeepSeek | 66.9 | 64 | 8 Oct 2026 |
| 11 |
gemini-2.0-pro-exp-02-05
|
65.1 | 74 | 8 Oct 2026 | |
| 12 | GD gemini-exp-1206 | Google DeepMind,Google | 64.1 | — | 8 Oct 2026 |
| 13 |
qwen2.5-max
|
Alibaba | 62.3 | — | 8 Oct 2026 |
| 14 | GD gemini-2.0-flash-001 | Google DeepMind,Google | 61.5 | 61 | 8 Oct 2026 |
| 15 |
DeepSeek: DeepSeek V3
|
DeepSeek | 60.5 | 62 | 8 Oct 2026 |
| 16 | GD gemini-2.0-flash-exp | Google DeepMind,Google | 59.3 | — | 8 Oct 2026 |
| 17 |
claude-3-5-sonnet-20241022
|
Anthropic | 59.0 | 45 | 8 Oct 2026 |
| 18 |
o1-mini
|
OpenAI | 57.8 | 65 | 8 Oct 2026 |
| 19 |
OpenAI: GPT-4o
|
OpenAI | 55.3 | 27 | 8 Oct 2026 |
| 20 |
DeepSeek: R1 Distill Llama 70B
💡 |
DeepSeek | 54.5 | 63 | 8 Oct 2026 |
| 21 |
grok-2-1212
|
xAI | 54.3 | 45 | 8 Oct 2026 |
| 22 |
gemini-2.0-flash-lite
|
54.3 | — | 8 Oct 2026 | |
| 23 |
gemini-2.0-flash-lite-preview-02-05
|
53.2 | — | 8 Oct 2026 | |
| 24 | D7 Dracarys2-72B-Instruct | Unknown | 52.6 | — | 8 Oct 2026 |
| 25 | L1 learnlm-1.5-pro-experimental | Unknown | 52.2 | — | 8 Oct 2026 |
| 26 |
Meta: Llama 3.3 70B Instruct
|
Meta | 50.2 | 34 | 8 Oct 2026 |
| 27 |
Google: Gemma 3 27B
|
50.0 | 47 | 8 Oct 2026 | |
| 28 |
claude-3-opus-20240229
|
Anthropic | 49.2 | 32 | 8 Oct 2026 |
| 29 |
mistral-large-2411
|
Mistral | 48.4 | 37 | 8 Oct 2026 |
| 30 |
Perplexity: Sonar
|
Perplexity | 46.9 | — | 8 Oct 2026 |
| 31 |
Qwen2.5 Coder 32B Instruct
|
Alibaba | 46.2 | — | 8 Oct 2026 |
| 32 | DL Dracarys2-Llama-3.1-70B-Instruct | Unknown | 46.2 | — | 8 Oct 2026 |
| 33 |
DeepSeek-R1-Distill-Qwen-32B
|
DeepSeek | 45.5 | 53 | 8 Oct 2026 |
| 34 |
mistral-small-2503
|
Mistral | 44.0 | 33 | 8 Oct 2026 |
| 35 | Az amazon.nova-pro-v1:0 | Amazon | 43.5 | — | 8 Oct 2026 |
| 36 |
claude-3-5-haiku-20241022
|
Anthropic | 43.5 | 30 | 8 Oct 2026 |
| 37 |
mistral-small-2501
|
Mistral | 42.5 | 31 | 8 Oct 2026 |
| 38 |
Microsoft: Phi 4
|
Microsoft | 41.6 | 41 | 8 Oct 2026 |
| 39 |
OpenAI: GPT-4o-mini
|
OpenAI | 41.3 | 31 | 8 Oct 2026 |
| 40 |
QwQ-32B-Preview
|
Alibaba | 40.2 | — | 8 Oct 2026 |
| 41 |
Google: Gemma 2 27B
|
38.2 | 21 | 8 Oct 2026 | |
| 42 | Az amazon.nova-lite-v1:0 | Amazon | 36.4 | — | 8 Oct 2026 |
| 43 | CC c4ai-command-r-plus-08-2024 | Cohere,Cohere for AI | 31.8 | — | 8 Oct 2026 |
| 44 | Az amazon.nova-micro-v1:0 | Amazon | 29.6 | — | 8 Oct 2026 |
| 45 |
gemma-2-9b-it
|
28.7 | 13 | 8 Oct 2026 | |
| 46 |
c4ai-command-r-08-2024
|
Cohere | 27.5 | — | 8 Oct 2026 |
| 47 |
Phi-3-small-8k-instruct
|
Microsoft | 24.0 | — | 8 Oct 2026 |
| 48 |
Phi-3-mini-4k-instruct
|
Microsoft | 22.4 | — | 8 Oct 2026 |
| 49 | AI OLMo-2-1124-13B-Instruct | Allen Institute for AI,University of Washington,New York University (NYU) | 22.1 | — | 8 Oct 2026 |
The Known Good
Provenance. These figures are published by LiveBench and redistributed here under Apache-2.0.
The publisher's own page is here. We did not run this evaluation and we do not adjust the published numbers — where a score feeds an index it is
min-max normalised against every other model holding the same evaluation, and nothing else is done to it.
This evaluation is one of the seven components of the
Known Good Index. Full licence detail is on attribution, and every score here is
in the CSV download.