About Ornith.site

Ornith.site is an independent, unofficial guide to Ornith 1.0, the open-source agentic coding model family released by DeepReinforce AI. We are not affiliated with, sponsored by, or endorsed by DeepReinforce AI.

Why This Site Exists

When Ornith 1.0 launched, information about it was scattered across research papers, model cards, GitHub threads, and social media. We built this site to bring it together in one place: verified benchmark numbers, a clear comparison of the four model sizes, working local-setup instructions, and honest comparisons against Claude, Qwen, and DeepSeek.

What We Cover

• Benchmarks — SWE-Bench, Terminal-Bench, ClawEval, and NL2Repo results, quoted from the official DeepReinforce evaluation with sources.
• Model comparison — 9B, 31B, 35B MoE, and 397B MoE: VRAM requirements, speed, and which one fits your hardware.
• Setup guides — tested instructions for vLLM, SGLang, Ollama, and LM Studio.
• Alternatives — how Ornith 1.0 compares to closed and open competitors on capability, price, and licensing.

How We Keep Information Accurate

Benchmark figures are taken from published evaluations and linked to their sources. Setup instructions are verified against the current model releases. When DeepReinforce ships new versions or updated numbers, we revise the affected pages and update the revision dates.

Contact

Spotted an error, or have a suggestion? We'd genuinely like to hear it — see our contact page or email hello@ornith.site.