nvcky/Ornith

VISIONConcurrent Unit Cost:1Model Size:9BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 12, 2026License:mitArchitecture:Transformer Open Weights Featherless Exclusive Cold

Ornith-1.0-9B is a 9 billion parameter model developed by DeepReinforce Team, part of the self-improving Ornith family of open-source models. It is specifically designed for agentic coding, excelling in benchmarks like Terminal-Bench 2.1, SWE-Bench, NL2Repo, and OpenClaw. This model utilizes a self-improving training framework with reinforcement learning to optimize both solution rollouts and the scaffolds that drive them, leading to higher-quality code generation and problem-solving.

Loading preview...

Ornith-1.0-9B: Agentic Coding Model

Ornith-1.0-9B, developed by the DeepReinforce Team, is a 9 billion parameter model optimized for agentic coding tasks. It is part of the Ornith family, which employs a unique self-improving training framework. This framework uses reinforcement learning to jointly optimize solution generation and the underlying scaffolds, enabling the model to discover better search trajectories and produce higher-quality solutions.

Key Capabilities & Features

  • State-of-the-Art Agentic Coding: Achieves strong performance among open-source models of comparable size on coding benchmarks such as Terminal-Bench 2.1 (e.g., 43.1 on Terminus-2, 40.6 on Claude Code), SWE-Bench (e.g., 69.4 Verified, 42.9 Pro), NL2Repo (27.2), and OpenClaw (63.1 Avg).
  • Self-Improving Training: Leverages an RL-based framework to enhance its ability to generate effective coding solutions and the reasoning processes behind them.
  • Reasoning Model: By default, the model generates a <think>...</think> block for chain-of-thought reasoning before providing the final answer, which can be parsed into a separate reasoning_content field.
  • Tool-Calling: Emits well-formed function calls that can be parsed into standard OpenAI-style tool_calls, making it compatible with various agent frameworks and coding CLIs.
  • Efficient Deployment: The 9B dense model is designed for efficient single-GPU deployment, requiring approximately 19 GB in bf16.
  • Open-Source License: Released under the MIT license, ensuring global accessibility and freedom from regional limitations.

Ideal Use Cases

  • Automated Code Generation: Generating Python functions, scripts, and solving coding challenges.
  • Agentic Development: Integrating with agent frameworks like Hermes, Atomic.chat, OpenClaw, and OpenHands for complex software engineering tasks.
  • Terminal-Based Coding Agents: Powering coding CLIs for understanding large codebases and automating development workflows.
  • Code Refactoring and Debugging: Utilizing its reasoning capabilities to analyze and suggest improvements or fixes in code.

Ornith-1.0-9B is a powerful choice for developers seeking an open-source model specialized in agentic coding with strong benchmark performance and efficient deployment.