v0.3.67

What's Changed

• Only disable cudnn on newer AMD GPUs. by @comfyanonymous in #10437
• Add custom node published subgraphs endpoint by @Kosinkadink in #10438
• execution: fold in dependency aware caching / Fix --cache-none with loops/lazy etc (Resubmit) by @rattus128 in #10440
• Small readme improvement. by @comfyanonymous in #10442
• WIP way to support multi multi dimensional latents. by @comfyanonymous in #10456
• Update template to 0.2.2 by @comfyui-wiki in #10461
• feat(api-nodes): network client v2: async ops, cancellation, downloads, refactor by @bigcat88 in #10390
• [V3] API Nodes: convert Tripo API nodes to V3 schema by @bigcat88 in #10469
• Remove useless function by @comfyanonymous in #10472
• [V3] convert Gemini API nodes to V3 schema by @bigcat88 in #10476
• Add warning for torch-directml usage by @comfyanonymous in #10482
• Fix mistake. by @comfyanonymous in #10484
• fix(api-nodes): random issues on Windows by capturing general OSError for retries by @bigcat88 in #10486
• Bump portable deps workflow to torch cu130 python 3.13.9 by @comfyanonymous in #10493
• Add a bat to run comfyui portable without api nodes. by @comfyanonymous in #10504
• Update template to 0.2.3 by @comfyui-wiki in #10503
• feat(api-nodes): add LTXV API nodes by @bigcat88 in #10496
• Update template to 0.2.4 by @comfyui-wiki in #10505
• frontend bump to 1.28.8 by @Kosinkadink in #10506

Full Changelog: v0.3.66...v0.3.67