Development

An ensemble of three labs' models hits 94.5% on ARC-AGI

development2026-02-05

Claim

A new state-of-the-art ARC-AGI submission reached 94.5% accuracy by ensembling GPT, Gemini and Claude, while China's open-weight trillion-parameter Intern-S1-Pro claimed state of the art on scientific reasoning and Kimi K2.5 set a new open-weight record.

Reported metric: 94.5%

Reporting date: 2026-02-05 (not necessarily the event date)

Sources

Attribution and reporting linked by this record; links do not imply independent verification.

Welcome to February 5, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

About

Intern-S1-Pro

Reported in

Welcome to February 5, 2026

Evidence for

Supersedes

Chinese models match the frontier on nineteen benchmarks

Referenced by

Cite or correct this record

Citation formatCitation text

An ensemble of three labs' models hits 94.5% on ARC-AGI
Innermost Loop KB — secondary record
Reporting date: 2026-02-05
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-02-05-arc-agi-945-by-ensemble
Claim: A new state-of-the-art ARC-AGI submission reached 94.5% accuracy by ensembling GPT, Gemini and Claude, while China's open-weight trillion-parameter Intern-S1-Pro claimed state of the art on scientific reasoning and Kimi K2.5 set a new open-weight record.
Attributed sources:
- Welcome to February 5, 2026 — https://theinnermostloop.substack.com/p/welcome-to-february-5-2026
This citation does not imply independent verification of linked-source claims.

Suggest a correction — opens a GitHub draft for you to review and submit.