Description
Ornith-1.5-35B-A3B is a sophisticated large language model that belongs to the Ornith-1.5 family. It is a mid-size mixture-of-experts model that activates approximately 3 billion parameters per token, providing a balance between performance and computational efficiency. This model is particularly notable for its ability to outperform its peers, such as Qwen 3.6-35B and dense models like Gemma 4-31B, across a variety of coding and agentic benchmarks.
Read More
The model extends the capabilities of its predecessor, Ornith-1.0, by expanding the self-improvement loop. This involves optimizing task generation, scaffold construction, and solution rollouts, allowing the model to continuously generate new training tasks and discover effective strategies for solving them. Reinforcement learning is employed to improve the policy, making Ornith-1.5-35B-A3B a dynamic and evolving tool.
Ornith-1.5-35B-A3B excels in various benchmarks, including Terminal-Bench 2.1, SWE-bench, and GPQA Diamond, showcasing its superior performance in coding and reasoning tasks. The model is designed to handle context windows of up to 262,144 tokens, with the capability to extend this limit using RoPE scaling, making it suitable for long-context applications.
The model is open-source and can be served using recent runtimes such as Transformers, vLLM, and SGLang. It supports OpenAI-compatible endpoints, allowing seamless integration with existing AI frameworks and tools. This makes Ornith-1.5-35B-A3B a versatile choice for developers and researchers looking to leverage advanced AI capabilities in their projects.
- CategoryAI Models & LLMs
- Pricing ModelOpen source
- DeploymentCloud
- PlatformsWeb
Ornith-1.5-35B-A3B Highlights
Mixture-of-experts model
3B activated parameters per token
Self-improvement loop
Reinforcement learning
Open-source
Supports long context windows
OpenAI-compatible endpoints
Superior benchmark performance
Getting Started with Ornith-1.5-35B-A3B
Access model page: Visit the Hugging Face model card
Authenticate: Ensure you have the necessary credentials
Configure: Set up the model with recommended parameters
Send prompts: Use the model for coding and reasoning tasks
Ornith-1.5-35B-A3B's Use Cases
- Advanced Coding
- Reasoning Tasks
- Agentic Coding
- Long-Context Applications
- AI Research
