tepirale/Ornith-Agents-A1-3.6-35B-A3B-task_arithmetic_v2

TEXT GENERATIONConcurrent Unit Cost:3Model Size:35.1BQuant:FP8Context Size:32kTool Calling:SupportedPublished:Jul 3, 2026License:apache-2.0Architecture:Transformer0.0K Open Weights Featherless Exclusive Cold

tepirale/Ornith-Agents-A1-3.6-35B-A3B-task_arithmetic_v2 is a 35.1 billion parameter language model merge created using the Task Arithmetic method, based on Qwen/Qwen3.5-35B-A3B. This model integrates capabilities from InternScience/Agents-A1 and deepreinforce-ai/Ornith-1.0-35B, specializing in advanced reasoning with a 32768 token context length. It is optimized for tool-calling and structured reasoning, leveraging Qwen3-style parsing for enhanced agentic behavior.

Loading preview...

Model Overview

tepirale/Ornith-Agents-A1-3.6-35B-A3B-task_arithmetic_v2 is a 35.1 billion parameter merged language model, built upon the Qwen/Qwen3.5-35B-A3B base using the Task Arithmetic merging method. This model combines the strengths of InternScience/Agents-A1 and deepreinforce-ai/Ornith-1.0-35B to enhance its agentic capabilities and reasoning.

Key Capabilities

  • Advanced Reasoning: Utilizes a qwen3 reasoning parser, enabling structured thought processes (e.g., <think>...</think>) for complex problem-solving.
  • Tool-Calling: Features a qwen3_xml tool-call parser, allowing for robust integration and execution of external tools.
  • Extended Context: Supports a substantial context length of 32768 tokens, facilitating the processing of lengthy inputs and maintaining conversational coherence.
  • Merge Method: Created with mergekit using Task Arithmetic, which combines models by adjusting weights based on their contributions to specific tasks.

Good For

  • Agentic Workflows: Ideal for applications requiring sophisticated agentic behavior, planning, and tool use.
  • Complex Problem Solving: Excels in tasks that benefit from explicit reasoning steps and structured output.
  • Research and Development: Provides a strong foundation for experimenting with merged models and agent-based AI systems.