Ornith-1.0 Hits 3M Downloads in Two Weeks
victormustar · x · 2026-07-14
Open-source model family Ornith-1.0 has surpassed 300 万 downloads on Hugging Face just two weeks after release.
Designed for agentic coding scenarios, it comes in various scales including 9B Dense、31B Dense、35B MoE、397B MoE, and reports scores across multiple coding benchmarks:
- Terminal-Bench 2.1：77.5
- SWE-Bench verified：82.4
- SWE-Bench pro：62.2
- SWE-Bench Multilingual：78.9
- NL2Repo：48.2
- SWE Atlas：41.2 / 42.6 / 39.1
- ClawEval：77.1
The authors state the model is post-trained on gemma4 and qwen3.5, utilizing a novel self-improvement training strategy: it generates not only problem-solving trajectories via reinforcement learning but also the task scaffolding driving these trajectories. The team plans to continue open-sourcing and improving the local LLM experience.
More from coding & agent
- AI agents are starting to strain code hosting platforms — craigsdennis · 2026-07-21
- Async OPD distillation doubles throughput while matching synchronous math accuracy — _lewtun · 2026-07-21
- Omnigent 0.6.0 adds Claude Code imports, Slack approvals and desktop apps — matei_zaharia · 2026-07-21
- Google appears to have quietly shipped Gemini 3.6 Flash, with lower pricing and better agentic scores — xiaohu · 2026-07-21
- Open-source CLI audits AI tools, MCP configs, and agent skills on local machines — Initial-Copy332 · 2026-07-21
- Coding agents feel less stressful when the 5-hour limits are temporarily removed — iamrobotbear · 2026-07-21