Loading page content
Historical Trends
The horizontal axis is model release date; the vertical axis is the highest overall score among models released by that point.Updated 09/10, 21:09
Current Open-weight leader
120 Open-weight models are included in the overall score
Kimi K3
Overall score 95.8
Current Proprietary leader
82 Proprietary models are included in the overall score
Claude Opus 5
Overall score 100.0
Highest Overall Score Over Time
Scores use the current Overall frontier domain's Bradley-Terry / Elo results, normalized so the leading model scores 100.
Highest open-weight scoreHighest proprietary score
Leadership Changes
Only dates when the leading open-weight or proprietary model changed are shown.
| Date | Open-weight leader | Proprietary leader |
|---|---|---|
| 2026-07 | Kimi K395.8 | Claude Opus 5100.0 |
| 2026-06 | GLM-5.286.7 | Claude Fable 598.8 |
| 2026-04 | Kimi-K2.682.7 | GPT-5.590.9 |
| 2026-03 | GPT-5.4 Pro88.8 | |
| 2026-02 | Claude Opus 4.684.6 | |
| 2026-01 | Kimi-K2.574.4 | GPT-5.2 Pro81.9 |
| 2025-12 | GLM-4.772.2 | GPT-5.2 Pro81.9 |
| 2025-11 | Gemini 3 Pro80.8 | |
| 2025-10 | GPT-5 Pro74.8 | |
| 2025-08 | GPT-573.2 | |
| 2025-07 | o371.9 | |
| 2025-04 | Qwen3-235B-A22B69.3 | o371.9 |
| 2025-03 | DeepSeek-R167.7 | |
| 2025-02 | DeepSeek-R167.7 | |
| 2025-01 | DeepSeek-R167.7 | Sonar Pro69.9 |
| 2024-12 | DeepSeek-V361.8 | OpenAI o167.7 |
| 2024-09 | Qwen2.5-72B59.8 | |
| 2024-07 | GPT-4o59.5 | |
| 2024-05 | DeepSeek-V2-050650.6 | GPT-4o59.5 |
| 2024-04 | Llama 3 8B Base43.9 | No model |
