Ornith 1.0 9B
deepreinforce-ai/Ornith-1.0-9B-GGUF
published Jun 2026 · updated Jun 2026
Ornith 1.0 9B is a text-generation model for agentic coding, trained with a self-improving reinforcement learning framework that jointly optimizes scaffolding and solution rollouts.
specs
| Task | Text Generation (Coding Agent) |
| Architecture | Dense Transformer |
| Parameters | 9 Billion |
| License | MIT |
about this model
The model employs a self-improving training framework based on reinforcement learning. It jointly optimizes both the solution rollout and the scaffold that drives the rollout, enabling the model to discover better search trajectories and generate higher-quality solutions. A three-layer defense mechanism prevents reward hacking by enforcing an immutable trust boundary, a deterministic monitor, and a frozen LLM judge.
Benchmark performance
On agentic coding benchmarks, Ornith-1.0-9B outperforms comparable models including Qwen3.5-9B, Qwen3.5-35B, Gemma4-12B, and Gemma4-31B across multiple evaluations:| Benchmark | Ornith-1.0-9B | Qwen3.5-9B | Qwen3.5-35B | Gemma4-12B | Gemma4-31B |
|---|---|---|---|---|---|
| Terminal-Bench 2.1 (Terminus-2) | 43.1 | 21.3 | 41.4 | 21 | 42.1 |
| Terminal-Bench 2.1 (Claude Code) | 40.6 | 18.9 | 38.9 | - | - |
| SWE-bench Verified | 69.4 | 53.2 | 70 | 44.2 | 52 |
| SWE-bench Pro | 42.9 | 31.3 | 44.6 | 27.6 | 35.7 |
| SWE-bench Multilingual | 52 | 39.7 | 60.3 | 32.5 | 51.7 |
| NL2Repo | 27.2 | 16.2 | 20.5 | 10.3 | 15.5 |
| Claw-eval Avg | 63.1 | 53.2 | 65.4 | 32.5 | 48.5 |
| SWE Atlas - QnA | 17.9 | 9.2 | 13.2 | - | - |
| SWE Atlas - RF | 16.6 | 4.3 | 10.2 | - | - |
| SWE Atlas - TW | 15.3 | 4.4 | 9.8 | - | - |
Ornith-1.0-9B is a reasoning model: by default the assistant response opens with a <think>…</think> block before the final answer. Released in June 2026 under the MIT license.best for
- ·Automated software engineering (SWE-bench tasks)
- ·Terminal-based coding agents
- ·Multi-language repository-level code generation
FAQ
It is best for agentic coding tasks such as resolving pull requests, terminal-based benchmarks, and multi-language software engineering.
With 9B parameters, it is designed for efficient single-GPU deployment and achieves state-of-the-art results among models of comparable size on coding benchmarks.
It is licensed under MIT, globally accessible and free from regional limitations.
It is a reasoning model that starts responses with a <think> block for chain-of-thought, then provides the final answer.
Use gigarouter's OpenAI-compatible endpoint with an API key to send prompts and receive generated text.
We're benchmarking and onboarding Ornith 1.0 9B as a hosted, OpenAI-compatible API. Sign in for free credit and be ready when it lands, or tell us you want it and we'll prioritize it.