Vals AI

Latest news

1. 11 Oct
   Agent teams waste tokens for little quality gain, benchmark finds

   Evals firm Vals AI tested solo versus team setups of GPT-6 Sol and Claude Opus 5.5 on the Vibe Code Bench and found teams cost 1.8x–5.1x more while delivering almost no consistent quality improvement. Anthropic’s internal tests with Opus 5.5 also showed diminishing returns as team size grew, with only modest score nudges beyond small teams.

   Research · 1 source

   

Articles

No articles about Vals AI yet.