LaunchAI ModelsAugust 20, 2026

Ornith-1.5 open-source LLM family matches Claude Opus 4.8

Ornith-1.5 spans 9B dense, 35B MoE, and 397B MoE, trained via self-improvement. The 397B scores 86.1 on Terminal-Bench 2.1 and 56.0 on DeepSWE, on par with Claude Opus 4.8 (85.0 and 59.0).

How this story unfolded

3 days · 4 reports · 10 community posts · 14 of 15 shown

  1. Aug 19
  2. Aug 20
  3. Aug 21
  4. Aug 23

Daily brief

Get tomorrow's AI brief in your inbox

More stories today

Open the live feed