tepirale/Ornith-Agents-A1-3.6-35B-A3B-dare_ties

TEXT GENERATIONConcurrency Cost:3Model Size:35.1BQuant:FP8Ctx Length:32kTool Calling:SupportedPublished:Jul 3, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Cold

tepirale/Ornith-Agents-A1-3.6-35B-A3B-dare_ties is a 35.1 billion parameter language model merged using the DARE TIES method, based on Qwen/Qwen3.5-35B-A3B. It integrates capabilities from deepreinforce-ai/Ornith-1.0-35B and InternScience/Agents-A1, specializing in advanced reasoning with a dedicated block and robust tool-calling functionalities. This model is optimized for complex problem-solving and agentic workflows, supporting a 32768 token context length.

Loading preview...

Model Overview

tepirale/Ornith-Agents-A1-3.6-35B-A3B-dare_ties is a 35.1 billion parameter merged language model, built upon the Qwen/Qwen3.5-35B-A3B base using the DARE TIES merging method. It combines the strengths of deepreinforce-ai/Ornith-1.0-35B and InternScience/Agents-A1 to deliver enhanced capabilities.

Key Capabilities

  • Advanced Reasoning: Features a dedicated <think>...</think> block for explicit reasoning processes, enabling more structured and transparent problem-solving.
  • Robust Tool-Calling: Optimized with qwen3_xml tool-call and reasoning parsers, making it highly effective for integrating with external tools and agentic workflows.
  • Extended Context: Supports a substantial context length of 32768 tokens, facilitating the processing of lengthy inputs and complex interactions.
  • Multimodal Potential: The underlying Qwen3.5 base and vLLM serving configuration suggest potential for multimodal inputs, specifically image processing, with a limit of 4 images per prompt.

Ideal Use Cases

  • Agentic AI Development: Excellent for building AI agents that require structured reasoning and interaction with external tools.
  • Complex Problem Solving: Suited for tasks demanding multi-step thought processes and logical deduction.
  • Code Generation & Analysis: Benefits from the reasoning capabilities for understanding and generating code, as demonstrated by its internal thought processes for terminal commands.
  • Interactive Applications: Its ability to handle long contexts and structured outputs makes it suitable for sophisticated conversational AI and interactive systems.

Finetunes
1 models