Meta AI5 min readSample
Open-weight models narrow the frontier gap on reasoning tasks
Publicly released weights are reported within a narrow band of closed frontier systems on deliberation-heavy evaluations.
mediumAGI relevance: mediummodelsbusiness
Agitology analysis
Why it matters
A narrowing gap changes who can run capable systems and who can audit them. Both matter for how quickly evidence about capability accumulates outside the labs producing it.
Key developments
- Gap on deliberation-heavy evaluations narrows to a small band
- Inference cost remains the practical differentiator
- Independent auditing of frontier-adjacent capability becomes feasible
Index impact
How this development moved — or failed to move — the dimensions it bears on.
- +0.3Reasoning