LocalLLaMA

Qwen 3.5 Tool Calling Fixes for Agentic Use: What’s Broken, What’s Fixed, What You (may) Still Need

Posted – What follows after this introduction is generated by Claude Opus 4.6 after hundreds of back and forths with log analysis for tool calls that were not working, and Qwen 3.5 models getting confused from local llm providers as well as Nano-Gpt. I…