LLM leaderboard
Updated November 16, 2026
Open the full page
Point estimates with Supported and Estimated evidence labels.
#
Model
Evidence
Score
1
Claude Fable 5.2
Anthropic
Estimated
Estimated
93.84
2
Gemini 4 Carbon
Google
Supported
Supported
89.56
3
Claude Opus 5.5
Anthropic
Supported
Supported
86.81
4
GPT-6 Astra
OpenAI
Supported
Supported
85.65
5
GPT-6.1 Sol
OpenAI
Supported
Supported
84.59
6
Claude Sonnet 5.5
Anthropic
Supported
Supported
83.13
7
Claude Fable 5.1
Anthropic
Supported
Supported
79.47
8
Gemini 4 Argon
Google
Supported
Supported
79.42
9
Gemini 4 Hydrogen
Google
Estimated
Estimated
79.65
10
GPT-6 Sol
OpenAI
Supported
Supported
77.39