Comfyui-Luck-gpt2.0 now supports gpt-image-2.5-flare / gpt-image-2.5-sunburst, and our docs page has been rewritten to match
Author luckdvr updated the plugin on 2026-09-10: the official node Comfyui-Luck gpt-image-2 adds both 2.5 models and their -2026-09-08 dated snapshots to the model (模型) dropdown, quality grows to six tiers (new xhigh / max), and the node takes up to 16 reference images plus a mask for inpainting. Node names, IDs and the default model gpt-image-2 are unchanged, so existing workflows do not switch on their own. When moving to 2.5, do not carry quality over unchanged: 2.5 high maps to the old medium, and only max maps to the old high.
The docs page had been stuck on April’s two-node version. It is now rewritten against the current plugin: the gpt-image-2-vip node, three prompt-control nodes (default gemini-3.5-flash), a five-model table, a 2.5 quality migration table and the new example workflow. Repo: github.com/luckdvr/Comfyui-Luck-gpt2.0.
📄 Docs: Luck GPT-Image 2 - ComfyUI Nodes
← Back to Live Updates · 📚 Monthly Archive