deepreinforce-ai/Ornith-1.0-9B

Ornith-1.0-9B: Agentic Coding Model

Ornith-1.0-9B, developed by DeepReinforce Team, is a 9 billion parameter model optimized for agentic coding. It is the most lightweight member of the Ornith-1.0 family, designed for efficient deployment on a single 80GB GPU.

Key Capabilities

• State-of-the-Art Agentic Coding: Achieves strong performance on coding benchmarks such as Terminal-Bench 2.1 (43.1 on Terminus-2, 40.6 on Claude Code), SWE-Bench (69.4 Verified, 42.9 Pro, 52 Multilingual), NL2Repo (27.2), and Claw-eval Avg (63.1).
• Self-Improving Training Framework: Employs Reinforcement Learning (RL) to jointly optimize solution rollouts and the underlying scaffolds, leading to improved search trajectories and higher-quality solutions.
• Reasoning Model: By default, the model generates a chain-of-thought within a <think>...</think> block before the final answer, which can be parsed into a separate reasoning_content field.
• Tool-Calling: Supports well-formed function calls, parsed into standard OpenAI-style tool_calls for seamless integration with agent frameworks.
• Open-Source License: Released under an MIT license, ensuring global accessibility and freedom from regional limitations.

Good For

• Agentic Coding Applications: Excels in scenarios requiring automated code generation, debugging, and task execution within a terminal environment.
• Integration with Agent Frameworks: Compatible with standard agent frameworks like Hermes, OpenClaw, and OpenHands due to its OpenAI-compatible endpoint and tool-calling capabilities.
• Local Development and Inference: Its 9B parameter size allows for comfortable deployment and serving on a single GPU, making it suitable for local development and rapid prototyping.
• Understanding Codebases and Automation: Optimized for tasks involving large codebases, automating repetitive coding tasks, and accelerating development workflows.