Open-source Ornith-1.5 release: 397B MoE beats Claude Opus 4.8 on four benchmarks
rohanpaul_ai · x · 2026-08-20
Ornith-1.5 is an open-source LLM family spanning 9B Dense, 35B MoE and 397B MoE, trained with self-improving strategies. The 397B model beats Claude Opus 4.8 on four benchmarks:
- Terminal-Bench 2.1: 86.1 vs 85
- SWE-bench Verified: 86 vs 85.8 (Pro 65.1, Multilingual 79.6)
- WideSearch: 80.8 vs 72.9
- BrowseComp: 86.6 vs 84.3
- Also: DeepSWE 56, HLE 44.6, ClawEval 81.4, Tool Decathlon 71.2
The architectural jump: task generation, agent design and solution generation live in the same RL loop — the model proposes its own tasks, writes the grading scaffold, produces solution rollouts, and propagates reward back through all three stages via GRPO. This self-generated curriculum searches near the model's capability frontier, removing much human task design.
More from Models
- Qwen3.8-Max tops frontend code leaderboard, beating Claude Fable 5 — Alibaba_Qwen · 2026-08-20
- Gemini 3.1 Pro Still the GOAT in Most Benchmarks Except Coding — Last_Conclusion_8984 · 2026-08-20
- Qwen3-Powered ASR Model superwhisper/s1-mini Trends on Hugging Face — superwhisper · 2026-08-20
- User tests reveal Qwen3.8-27B knowledge regression vs 3.6 — EmPips · 2026-08-20
- Qwen3.8 27B Speeds Up 3x Post-Release: Open Source Advantage — TheMoonMidas · 2026-08-20
- Mac can now run a 27B model locally that codes, reasons, and sees — TheMoonMidas · 2026-08-20