Custom ComfyUI Node Boosts Qwen-Image 2.1 Prompt Enhancer 1.4-1.7x via MTP Head
mozophe · reddit · 2026-09-25
A developer found ComfyUI's official Qwen-Image 2.1 prompt enhancer (a Qwen3.5-9B fine-tune) slow due to a missing MTP head. Attaching an MTP head boosted throughput on an RTX 4090 Laptop 16GB: 35→47 tok/s for t2i (1.36x) and 31→52 tok/s for editing (1.65x); vs the official maxlength checkpoint, up to 1.7x faster for t2i and 2.5x for editing, with unchanged quality (same seed yields different text).
Since ComfyUI needed MTP-specific fixes for Qwen 3.5 image input, the author released a custom node: ComfyUI-Qwen-Image-2.1-PromptEnhancer-MTP.
Details:
- Requires ComfyUI v0.37.0+, 16GB VRAM recommended (peak 14GB t2i, 16GB editing), 32GB RAM, 10GB disk per model
- Auto-downloads and attaches the MTP head on first use; uses Qwen's official system prompts and sampling settings
- Includes heretic (uncensored) versions with the same speed-up
- Supports editing with up to 10 input images; prompt and reasoning output separately; sample workflows included
More from Infra
- UK Sovereign AI invests in Fractile to build next-gen frontier AI chips — charlieharris01 · 2026-11-11
- What happened when I moved vector retrieval into Snowflake SQL instead of orchestrating RAG outside — dima806_dima · 2026-11-11
- AI chip startup Fractile raises $720M Series C for DRAM-on-processor design — Tom_Westgarth15 · 2026-11-11
- Mobile AI Factory: agentic pipeline builds on-device SLMs that beat 5-shot frontier models — kalyan_kpl · 2026-11-11
- Vivo's Starlink direct-to-cell service hits 10M users in two weeks across Brazil — XFreeze · 2026-11-11
- Nan Ya Plastics to spend NT$11.53B on fifth glass fabric plant on AI server demand — zephyr_z9 · 2026-11-11