Skip to main content
← Back to sources

DeepReinforce Releases Ornith-1.0 — MIT-Licensed "Self-Scaffolding" Agentic Coding Models (9B–397B MoE) Built on Gemma 4 and Qwen 3.5

Published 2026-06-29Ingested 2026-06-30Foundation ModelsMedium⭐ Timeline Candidate

Summary

DeepReinforce released Ornith-1.0 on June 29, 2026, an open-weights model family aimed at agentic coding, under an MIT license. The lineup spans a 9B dense, a 31B dense, a 35B mixture-of-experts, and a 397B MoE variant, fine-tuned on top of Gemma 4 and Qwen 3.5 (both Apache 2.0) bases. DeepReinforce claims state-of-the-art performance among open-source models of comparable size on coding benchmarks. The "self-scaffolding" framing refers to the model's ability to autonomously manage multi-step to

Alignment: New signal not yet covered
Related Positions: Multi-Model, Multi-Vendor Strategy, AI-Assisted Development Tooling, Agentic Workflows
open-source-modelsagentic-codingdeepreinforceornithmixture-of-expertsqwengemmaself-hostingopen-weights