[Bug] Agentic reasoning produces inferior outputs compared to alternative models on fact-checking tasks

Open 💬 0 comments Opened Jul 10, 2026 by neoncapy

Bug Description
Fable here is doing a worse job over and over than Sol from chatgbt with chatgbt's Sol having to correct it at fact checking an email

Environment Info

  • Platform: darwin
  • Terminal: vscode
  • Version: 2.1.206
  • Feedback ID: c956ac67-64b8-456e-92fd-27a27b8a0d18

Errors

[]

View original on GitHub ↗