Ornith-1.5 Open Models Claim Claude Opus-Level Performance

Ornith-1.5 open-source models launched in 9B Dense, 35B MoE and 397B MoE variants, trained with a self-improvement loop where models generate their own tasks. The team claims Claude Opus-level performance, surpassing Claude on Terminal-Bench.

2026-08-26 ~ 2026-08-27 · 2 related posts

• Ornith-1.5 Open Models Released, Claiming Claude Opus Performance — alejandroll10 · 2026-08-26
• Ornith-1.5: AI Generates Its Own Training Data, Beats Claude on Terminal-Bench — 新智元 · 2026-08-27