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.

Closed weights (best to date)Open weights (best to date)

GPT-6 AstraOpenAI · released 3 Sep 2026ECI 169.1Closed weights

Data: Epoch AI, Capabilities & benchmarking (CC BY 4.0), last fetched 5 Nov 2026. A capability index, not real-world usefulness or adoption. Download the chart data (CSV)

ECI with 90% interval. The top 10 overlap with the leader: treat it as a tie, not a winner.

MODEL	ECI	RANGE
GPT-6 AstraOpenAI · closed	169.1	166.2–172.3
GPT-6.1 SolOpenAI · closed	168.9	166.0–172.2
Claude Opus 5.5Anthropic · closed	168.4	165.5–171.5
GPT-6.1 TerraOpenAI · closed	167.8	164.9–171.2
Claude Fable 5.1Anthropic · closed	166.9	164.3–169.8
Gemini 4 ArgonGoogle DeepMind · other	166.4	163.8–169.4
Claude Sonnet 5.5Anthropic · closed	166.2	163.5–169.2
Claude Opus 5Anthropic · closed	164.4	162.0–167.2
DeepSeek V4.1 Flash 1017DeepSeek · open	159.3	157.0–161.8
Kimi K3Moonshot · open	158.5	156.5–160.6

9.8points between the best closed and best open model

+19.1closed frontier gain since GPT-5 (Aug 2025)

Best model by lab

Highest ECI per organisation in Epoch’s data.

Distinct publishers mentioning each tool, 20 Oct 2026–26 Oct 2026. Attention, not adoption.

Epoch’s Capabilities Index combines many public benchmarks into one score so models can be compared over time.

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.

Fetched every morning. If a fetch fails we keep the last good copy and show its date. Last successful fetch: 5 Nov 2026.