Current runtime candidate
Know what your workflow can use.
A complete view of the models, LoRAs and custom-node repositories currently pinned for the three Comfy Rail runtime profiles.
This is the current pinned candidate inventory from runtime/models.lock.json and runtime/custom-nodes.lock.json. It is not a commercial release-readiness or automatic workflow-compatibility guarantee.
01 · Models
Pinned model components
Classes show how ComfyUI uses each component. Runtime labels show where the file is currently installed.
| Component | Class | Source | Runtime |
|---|---|---|---|
qwen_3_4b.safetensorsclip/qwen_3_4b.safetensors | Text encoder | Comfy-Org/z_image_turbo | GenerateEditUpscale |
z_image_turbo_bf16.safetensorsdiffusion_models/z_image_turbo_bf16.safetensors | Diffusion model | Comfy-Org/z_image_turbo | GenerateEditUpscale |
vae_ultraflux.safetensorsvae/vae_ultraflux.safetensors | VAE | Owen777/UltraFlux-v1 | GenerateEditUpscale |
4x NMKD Superscaleupscale_models/4x_NMKD-Superscale-SP_178000_G.pth | Upscale model | OpenModelDB/4x-NMKD-Superscale | GenerateEditUpscale |
Stable Diffusion safety checkercustom_node_assets/comfyui-tooling-nodes/safetychecker/model.safetensors | Safety classifier | CompVis/stable-diffusion-v1-4 | GenerateEditUpscale |
Qwen3-4B-Instruct-2507LLM/Qwen3-4B-Instruct-2507/ | Language model | Qwen/Qwen3-4B-Instruct-2507 | GenerateUpscale |
DA3 Metric Largedepthanything3/da3metric_large.safetensors | Depth estimation | depth-anything/DA3METRIC-LARGE | EditUpscale |
Z-Image Turbo Fun ControlNet Union 2.1model_patches/Z-Image-Turbo-Fun-Controlnet-Union-2.1.safetensors | ControlNet | alibaba-pai/Z-Image-Turbo-Fun-Controlnet-Union-2.1 | EditUpscale |
Qwen 2.5 VL 7B FP8clip/qwen_2.5_vl_7b_fp8_scaled.safetensors | Vision-language encoder | Comfy-Org/Qwen-Image_ComfyUI | Edit |
Qwen Image VAEvae/qwen_image_vae.safetensors | VAE | Comfy-Org/Qwen-Image_ComfyUI | Edit |
Qwen Image Q3 K Mdiffusion_models/qwen-image-Q3_K_M.gguf | Quantized diffusion model | city96/Qwen-Image-gguf | Edit |
Qwen Image InstantX Inpaintingcontrolnet/Qwen-Image-InstantX-ControlNet-Inpainting.safetensors | Inpainting ControlNet | Comfy-Org/Qwen-Image-InstantX-ControlNets | Edit |
Qwen3-VL-4B-InstructLLM/Qwen3-VL-4B-Instruct/ | Vision-language model | Qwen/Qwen3-VL-4B-Instruct | Edit |
FlashVSR Streaming DMDFlashVSR-v1.1/diffusion_pytorch_model_streaming_dmd.safetensors | Video super-resolution | JunhaoZhuang/FlashVSR-v1.1 | Upscale |
FlashVSR LQ ProjectionFlashVSR-v1.1/LQ_proj_in.ckpt | Projection model | JunhaoZhuang/FlashVSR-v1.1 | Upscale |
FlashVSR Temporal DecoderFlashVSR-v1.1/TCDecoder.ckpt | Temporal decoder | JunhaoZhuang/FlashVSR-v1.1 | Upscale |
Wan 2.1 VAEFlashVSR-v1.1/Wan2.1_VAE.pth | VAE | JunhaoZhuang/FlashVSR-v1.1 | Upscale |
02 · LoRAs
Pinned LoRA components
LoRA filenames are especially likely to be renamed in existing workflows. Match by purpose and source before mapping the installed path.
| Component | Class | Source | Runtime |
|---|---|---|---|
Qwen Image Lightning 8 Steps V1.1loras/Qwen-Image-Lightning-8steps-V1.1.safetensors | Acceleration LoRA | lightx2v/Qwen-Image-Lightning | Edit |
HBGv4 Full 4250loras/z-image model HBGv4 FUll 4250.safetensors | First-party style LoRA | 3DOfficeAT e.U. | GenerateEditUpscale |
03 · Custom Nodes
Pinned custom-node repositories
All three runtime candidates install these exact repositories at pinned commits. Core ComfyUI nodes are available in addition to this list.
| Custom node | Class | Repository | Pinned commit |
|---|---|---|---|
ComfyUI-Custom-ScriptsCN-003 | Workflow UI utilities | pythongosssss/ComfyUI-Custom-Scripts | f2838ed5 |
ComfyUI_LayerStyleCN-004 | Layer compositing | chflame163/ComfyUI_LayerStyle | 3b17c7f4 |
ComfyUI-KJNodesCN-007 | Image and workflow utilities | kijai/ComfyUI-KJNodes | 79f529a8 |
ComfyUI_UltimateSDUpscaleCN-008 | Tiled diffusion upscaling | ssitu/ComfyUI_UltimateSDUpscale | d6b575ad |
ComfyUI-Inpaint-CropAndStitchCN-011 | Inpainting workflow | lquesada/ComfyUI-Inpaint-CropAndStitch | a5fbe06f |
ComfyUI-FlashVSR_Ultra_FastCN-013 | Video super-resolution | lihaoyun6/ComfyUI-FlashVSR_Ultra_Fast | 4820b3f0 |
comfyui-propostCN-016 | Post-processing | digitaljohn/comfyui-propost | df6a6d12 |
ControlAltAI-NodesCN-017 | Control and image utilities | gseth/ControlAltAI-Nodes | 721492b6 |
ComfyUI_QwenVLCN-019 | Vision-language | alexcong/ComfyUI_QwenVL | 56f28865 |
ComfyUI-VAE-UtilsCN-021 | VAE utilities | spacepxl/ComfyUI-VAE-Utils | a4886787 |
ComfyUI-GGUFCN-022 | Quantized model loader | city96/ComfyUI-GGUF | 795e4515 |
ComfyUI-Detail-DaemonCN-023 | Sampling and detail control | Jonseed/ComfyUI-Detail-Daemon | 39206d10 |
ComfyUI-DepthAnythingV3CN-024 | Depth estimation | PozzettiAndrea/ComfyUI-DepthAnythingV3 | 20053ca1 |
comfyui-tooling-nodesCN-025 | Safety and workflow tooling | Acly/comfyui-tooling-nodes | 7fc3df11 |
Need a precise answer?