Build Failed with any official Qwen LLM (Qwen3.5-2B, 4B) on M1

[nailarch]

mtplx doctor --json

Build failed
Warning: You are sending unauthenticated requests to the HF Hub. Please set a HF_TOKEN to enable higher rate limits and faster downloads.
[forge] downloading Qwen/Qwen3.5-2B [forge] source format: bf16_native [forge] converting with mlx-Im
[forge] restored vision tower: 297 tensors (662833152 bytes) in model-vision.safetensors...

Exact command

MTPLX

Model path or repo id

Qwen/Qwen3.5-2B

Chip, RAM, macOS version

M1, 8G, Tahoe 26.6

[youssofal]

Need the actual failure to act on this. Post the full build output, mtplx doctor --json, and how much RAM the M1 has, official small Qwen models are exercised regularly so the interesting part is what is different on your machine.

[nailarch]

I think problem appeared after installing and removing hermes agent using BREW.
I'll write later if I find out exactly why this happened.