I encountered some issues when using llama-3.3-70b-versatile and qwen/qwen3-32b with the OpenCode agent. The two models that are good for coding fail to correctly use the simple file editing tools: {“message”:“Failed to call a function. Please adjust your prompt. See ‘failed_generation’ for more details.”,“type”:“invalid_request_error”}. Only the GPT OSS 120B model successfully uses the tools, but it’s not good at all for coding. I hope Groq fixes these problems ASAP so we can enjoy Vibe coding with the fastest inference.