aiminute. ← All AI news
Research AI Minute Newsroom 2026-10-04

Pretraining on everyday photos got a model to 70% on a reasoning test.

Pretraining on everyday photos got a model to 70% on a reasoning test.

Kaiming He's MIT team built a vision-only solver for ARC, a grid puzzle test of abstract reasoning. Starting from a model pretrained on natural images, an ensemble scored 70.2% where a specialised language system scored 71.6%. Frontier models still win easily, with Gemini 3 Pro at 98% on the same benchmark.

Why it mattersAbstract reasoning may be partly a seeing problem, not only a language one. That points to much smaller models for tasks we assumed needed giant ones.

✓ Verified · 3 sources

WhatsApp X Telegram
Read in the app — free, in 9 languages

Related stories

The IEA expects $4 trillion more in data center spending this decade.
2026-11-19
OpenAI says its models improved AMD's flagship graphics chip design.
2026-11-19
OpenAI says agents finish 46% of its researchers' half-day tasks.
2026-11-19
Four security teams used AI to find the same critical Nginx flaws.
2026-11-19
Indeed study finds neither AI job losses nor a productivity boom.
2026-11-18