Intelligence vs Time
Every ranked LLM plotted by intelligence index against its release date, with the best-so-far line, open vs closed weights and the yearly pace.
Best so far: 93.7, GPT-6 Astra (2026/09/03). Open weights: 84.4, Kimi K3 (2026/07/16); closed weights: 93.7; a gap of 9.3 points. Over the last 12 months (Nov 2025 – Oct 2026) the best so far rose at +21.7 index points a year (a least-squares line through its value at each month's end).
| Model | Released | Weights | Index | Gain |
|---|---|---|---|---|
| GPT-6 Astra OpenAI | 2026/09/03 | Closed | 93.7 | +2.0 |
| Claude Fable 5.1 Anthropic | 2026/09/01 | Closed | 91.7 | +0.5 |
| Claude Fable 5 superseded Anthropic | 2026/06/09 | Closed | 91.2 | +5.5 |
| Claude Opus 4.8 superseded Anthropic | 2026/05/27 | Closed | 85.7 | +0.7 |
| Gemini 3.1 Pro Google | 2026/02 | Closed | 85.0 | +7.6 |
| GPT-5.2 Pro superseded OpenAI | 2025/12/10 | Closed | 77.4 | +2.5 |
| GPT-5.1-Codex-Max OpenAI | 2025/11/19 | Closed | 74.9 | +0.6 |
| Qwen3-VL-235B-A22B-Thinking superseded Alibaba | 2025/05/14 | Open | 74.3 | +10.5 |
| gemini-2.5-pro-preview-03-25 Google | 2025/05/07 | Closed | 63.8 | +3.8 |
| Phi-4-reasoning Microsoft | 2025/04 | Open | 60.0 | +6.4 |
| Gemini 2.5 Pro Google | 2025/03 | Closed | 53.6 | +2.5 |
| OpenAI o1 superseded OpenAI | 2024/12/17 | Closed | 51.1 | +18.3 |
| Qwen2.5 72B Instruct Alibaba | 2024/09/19 | Open | 32.8 | +4.4 |
| Qwen2.5-32B-Instruct superseded Alibaba | 2024/09 | Open | 28.4 | +5.2 |
| Llama-3.1-405B-Instruct Meta | 2024/07/23 | Open | 23.2 | +2.2 |
| GPT-4 Turbo OpenAI | 2024/04/09 | Closed | 21.0 | +10.2 |
| GPT-4 OpenAI | 2023/05/28 | Closed | 10.8 | — |