Development

A harness hits 99% on a benchmark with zero weight changes

development2026-07-17

Claim

The Schema harness, which has models write each game's mechanics as an executable program, hit roughly 99% on ARC-AGI-3 with zero weight changes, proof that scaffolding is still free intelligence.

Reported metric: ~99%, no weight changes

Reporting date: 2026-07-17 (not necessarily the event date)

Sources

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

Welcome to July 17, 2026 (theinnermostloop.substack.com)

Knowledge graphopen in browser →

Reported in

Welcome to July 17, 2026

Evidence for

Supersedes

A model beats an ARC-AGI-3 game and hits 92.5% on ARC-AGI-2

Referenced by

Cite or correct this record

Citation formatCitation text

A harness hits 99% on a benchmark with zero weight changes
Innermost Loop KB — secondary record
Reporting date: 2026-07-17
KB permalink: https://nicholsn.github.io/innermost-loop-kb/developments/2026-07-17-scaffolding-is-still-free-intelligence
Claim: The Schema harness, which has models write each game's mechanics as an executable program, hit roughly 99% on ARC-AGI-3 with zero weight changes, proof that scaffolding is still free intelligence.
Attributed sources:
- Welcome to July 17, 2026 — https://theinnermostloop.substack.com/p/welcome-to-july-17-2026
This citation does not imply independent verification of linked-source claims.

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