[Bug]: 22GB VRAM usage for 0.6B Qwen3-TTS model (2-stage architecture overhead)
March 30, 2026 · #2318
Python
Difficulty: Medium
Labels
bug
Parent Repository
vllm-project/vllm-omni
Python repository
4,212 730
Labels
vllm-project/vllm-omni
Python repository
Sign in required
Authenticate to use favourites & bookmarks