The Pulse
Where AI models stand today
Best score to date on the Epoch Capabilities Index, a composite of many benchmarks. Select a step to see the model that set it.
ECI 173.4Closed weightsMonth by month
Pick a month to see its leaders, releases and movement. The table shows the frontier score, the best open model and the gap at each month's end. Scores: Epoch Capabilities Index (CC BY 4.0).
| MONTH | FRONTIER | MODELS | BEST OPEN | GAP | Δ GAP | RELEASES |
|---|---|---|---|---|---|---|
| 2026-11 | 173.4 | Claude Fable 5.2Kimi K3.1 Pro (open) | 161.2 | 12.2 | +4.2 | 4 |
| 2026-10 | 169.2 | GPT-6 AstraKimi K3.1 Pro (open) | 161.2 | 8.0 | -2.6 | 10 |
| 2026-09 | 169.2 | GPT-6 AstraKimi K3 (open) | 158.6 | 10.6 | +4.5 | 12 |
| 2026-08 | 164.7 | Claude Opus 5Kimi K3 (open) | 158.6 | 6.1 | 0.0 | 8 |
| 2026-07 | 164.7 | Claude Opus 5Kimi K3 (open) | 158.6 | 6.1 | -5.0 | 12 |
| 2026-06 | 164.2 | Claude Fable 5GLM-5.2 (open) | 153.1 | 11.1 | -0.1 | 7 |
| 2026-05 | 162.8 | GPT-5.5 ProKimi K2.6 (open) | 151.6 | 11.2 | 0.0 | 3 |
| 2026-04 | 162.8 | GPT-5.5 ProKimi K2.6 (open) | 151.6 | 11.2 | -0.3 | 16 |
| 2026-03 | 160.3 | GPT-5.4 ProKimi K2.5 (open) | 148.8 | 11.5 | +2.9 | 6 |
| 2026-02 | 157.3 | GPT-5.2 ProKimi K2.5 (open) | 148.8 | 8.6 | 0.0 | 10 |
| 2026-01 | 157.3 | GPT-5.2 ProKimi K2.5 (open) | 148.8 | 8.6 | -2.5 | 1 |
| 2025-12 | 157.3 | GPT-5.2 ProDeepSeek-V3.2 (open) | 146.3 | 11.1 | +3.5 | 5 |
| 2025-11 | 153.0 | Gemini 3 ProKimi K2 Thinking (open) | 145.5 | 7.6 | +0.7 | 4 |
| 2025-10 | 151.3 | GPT-5 ProDeepSeek-V3.2-Exp (open) | 144.4 | 6.9 | +1.3 | 2 |
| 2025-09 | 150.0 | GPT-5DeepSeek-V3.2-Exp (open) | 144.4 | 5.6 | -1.2 | 5 |
| 2025-08 | 150.0 | GPT-5Qwen3-235B-A22B-Thinking (Jul 2025) (open) | 143.2 | 6.8 | +2.6 | 7 |
| 2025-07 | 147.5 | o3-proQwen3-235B-A22B-Thinking (Jul 2025) (open) | 143.2 | 4.2 | -2.5 | 6 |
| 2025-06 | 147.5 | o3-proDeepSeek-R1 (May 2025) (open) | 140.7 | 6.7 | +0.4 | 5 |
| 2025-05 | 147.0 | o3DeepSeek-R1 (May 2025) (open) | 140.7 | 6.3 | -2.1 | 6 |
| 2025-04 | 147.0 | o3DeepSeek-R1 (open) | 138.6 | 8.4 | +3.0 | 14 |
| 2025-03 | 144.0 | Gemini 2.5 Pro (Mar 2025)DeepSeek-R1 (open) | 138.6 | 5.4 | +2.9 | 4 |
| 2025-02 | 141.2 | o1DeepSeek-R1 (open) | 138.6 | 2.5 | 0.0 | 4 |
| 2025-01 | 141.2 | o1DeepSeek-R1 (open) | 138.6 | 2.5 | -7.0 | 4 |
| 2024-12 | 141.2 | o1DeepSeek-V3 (open) | 131.7 | 9.5 | +2.1 | 5 |
| 2024-11 | 135.3 | o1-miniLlama 3.1-405B (open) | 127.9 | 7.4 | 0.0 | 2 |
| 2024-10 | 135.3 | o1-miniLlama 3.1-405B (open) | 127.9 | 7.4 | 0.0 | 2 |
| 2024-09 | 135.3 | o1-miniLlama 3.1-405B (open) | 127.9 | 7.4 | +5.3 | 12 |
| 2024-08 | 130.0 | Claude 3.5 SonnetLlama 3.1-405B (open) | 127.9 | 2.1 | 0.0 | 1 |
| 2024-07 | 130.0 | Claude 3.5 SonnetLlama 3.1-405B (open) | 127.9 | 2.1 | -5.6 | 6 |
| 2024-06 | 130.0 | Claude 3.5 SonnetQwen2-72B (open) | 122.3 | 7.7 | +0.9 | 5 |
| 2024-05 | 128.1 | GPT-4o (May 2024)Llama 3-70B (open) | 121.3 | 6.8 | +2.3 | 5 |
| 2024-04 | 125.8 | GPT-4 Turbo (Apr 2024)Llama 3-70B (open) | 121.3 | 4.5 | -7.9 | 8 |
| 2024-03 | 125.6 | Claude 3 OpusMixtral 8x7B (open) | 113.2 | 12.4 | 0.0 | 2 |
| 2024-02 | 125.6 | Claude 3 OpusMixtral 8x7B (open) | 113.2 | 12.4 | +0.8 | 9 |
| 2024-01 | 124.8 | GPT-4 Turbo (Nov 2023)Mixtral 8x7B (open) | 113.2 | 11.6 | 2 |
Best model by lab
Highest ECI per organisation in Epoch’s data.
| LAB | MODEL | ECI |
|---|---|---|
| Anthropic | Claude Fable 5.29 Nov 2026 · closed | 173.4 |
| OpenAI | GPT-6.2 Astra17 Nov 2026 · closed | 172.8 |
| Google DeepMind | Gemini 4 Carbon10 Nov 2026 · closed | 169.7 |
| Meta AI | Muse Wonder12 Oct 2026 · closed | 161.7 |
| Moonshot | Kimi K3.1 Pro23 Oct 2026 · open | 161.2 |
| DeepSeek | DeepSeek V4.1 Flash 101717 Oct 2026 · open | 159.4 |
| xAI | Grok 4.612 Aug 2026 · closed | 157.8 |
| Alibaba | Qwen 3.8 Max2 Aug 2026 · closed | 157.6 |
| Z.ai (Zhipu AI) | GLM-5.314 Aug 2026 · open | 157.1 |
| Mistral AI | Mistral Large 429 Oct 2026 · open | 154.6 |
Source: Epoch AI, Capabilities Index (CC BY 4.0).
Tools builders are discussing
All tools →Distinct publishers mentioning each tool, 16 Nov 2026–22 Nov 2026. Attention, not adoption.
Claude94Claude Code29ChatGPT20Gemini16Higgsfield7Codex7n8n5DeepSeek4How to read the Pulse
What the index is
Epoch’s Capabilities Index combines many public benchmarks into one score so models can be compared over time.
What it is not
It does not measure cost, speed, reliability on your task, or how many people use a model. Check the speedruns and our own tests for that.
Freshness
Updated each time the morning run collects data. If a fetch fails we keep the last good copy and show its date. Last successful fetch: 22 Nov 2026.