K2 Horizon declines rather than guesses, cutting hallucination to 26%
ArtificialAnlys · x · 2026-09-03
Artificial Analysis notes K2 Horizon 375B A23B attempts just 40% of AA-Omniscience questions and declines the rest rather than guessing, producing a 26% hallucination rate — among the lowest measured — with accuracy of 18% essentially unchanged. This is a huge improvement over K2 Think V2's 71%, moving the AA-Omniscience Index from -40 to -3.
Related event: K2 Horizon Evaluation Details Show 26% Hallucination Rate, Agent Lead(4 posts)→
More from Models
- GPT-6 Astra and Claude Fable 5.1 write $10,000 research papers — experts reject all — sayashk · 2026-10-21
- GitHub Copilot's HydraFusion Orchestrates Multiple Models Per Task — WirelessLife · 2026-10-21
- OpenVLHarness turns any MLLM into a visual reasoner, +9.9 points across 22 datasets — yining_hong · 2026-10-21
- User claims Gemini analyzed his tennis match video and improved his game — zacharynado · 2026-10-21
- Opus 5.5 tops OpenRouter spending in 20 of 29 task categories — mmurph · 2026-10-21
- Meta team builds open-source 30B LLM from scratch in months, hails 'second era' — TimDarcet · 2026-10-21