Overview

teolm30/fox1.2 is a highly efficient, fine-tuned Qwen2.5-0.5B model, developed by teolm30, specifically engineered for seamless integration and execution within the OpenClaw agent framework. With approximately 494 million parameters and a substantial 32,768-token context length, this model is designed for local inference on consumer hardware, requiring only about 6GB of VRAM. Its core strength lies in its 100% support for all OpenClaw agent tools, enabling robust automation and task execution.

Key Capabilities

• Comprehensive Tool Support: Fully supports all OpenClaw tools, including exec (shell commands), read/write/edit (file operations), web_search/web_fetch (web interactions), process (background session management), cron (job management), and memory_search/memory_get (memory operations).
• Smart Tool Selection: Optimized to intelligently select and generate appropriate tool calls based on user prompts, streamlining agent workflows.
• Compact & Efficient: At ~494M parameters and ~994MB (F16), it's built for fast, local inference, making it accessible for a wide range of hardware.
• Dedicated Training: Trained on over 200 examples covering diverse OpenClaw tool patterns to ensure high accuracy in tool call generation.

Good For

• Developers building OpenClaw agents requiring reliable and efficient tool execution.
• Automated scripting and task management where shell commands, file system access, and web interactions are crucial.
• Applications needing a compact, fast, and locally runnable model for agentic workflows.
• Scenarios demanding smart tool selection for complex multi-step operations.