ComfyUI · 2026-09-09 · notable
ComfyUI v0.35.0 — a Comfy Compiler plus GPT-6 Astra and Fable 5.1 nodes
ComfyUI v0.35.0 introduces the Comfy Compiler and adds nodes for OpenAI GPT-6 Astra, GPT Image 2.5, Claude Fable 5.1, Google Omni 1.1 and Meta Muse Image. A File3DToMesh node parses GLB, GLTF, OBJ and STL files into meshes.
ComfyUI's September release wires in the newest frontier image and video models and adds a compiler that handles memory.
What is it?
The Comfy Compiler arrives in ComfyUI v0.35.0, alongside partner nodes for OpenAI GPT-6 Astra and GPT Image 2.5, Claude Fable 5.1, Google Omni 1.1, MiniMax-H3 Max Turbo, Recraft V4 Styles Pro and Meta Muse Image. On the 3D side, a File3DToMesh node parses GLB, GLTF, OBJ and STL files into a MESH, and Pixal3D multiview models are supported.
How does it work?
The release notes describe the compiler through its guardrails rather than its design — it pauses for long-lived sparse allocations, and several fixes in this version target memory handling. The rest of the update is node work: VideoTrim and VideoCrop gain editing widgets, Save Image Advanced writes AVIF, HEVC remuxing to MP4 and MOV improves, and PyTorch SDPA attention is enabled on more AMD GPUs.
Why does it matter?
ComfyUI is where much open-source image and video work actually runs, so how quickly a new model gets a node decides how fast it reaches those graphs. This release also cleans up alpha-channel handling across several nodes, adds Porter-Duff blend modes with transparency and a colour-space conversion node, and improves memory management for container deployments.
Who is it for?
diffusion and generative-video artists