Loading
Du hast einen ComfyUI-Workflow? Erstelle direkt im Browser eine verständliche Workflow-Dokumentation.Workflow verständlich machen.Jetzt ausprobieren ↗
If you ever downloaded the H3 GGUF quants and got an error the instant ComfyUI touched the file, this is the fix — and the error was never your fault or the quant's.
Laufzeitprofil
Quellenbeschreibung
If you ever downloaded the H3 GGUF quants and got an error the instant ComfyUI touched the file, this is the fix — and the error was never your fault or the quant's.
GGUF models now load with no manual step
ComfyUI-GGUF validates a GGUF's architecture against a fixed list and rejects anything not on it before reading a single tensor . Upstream's list has no minimax_h3 entry, so every MiniMax-H3 DiT quant died with:
Unexpected architecture type in GGUF file: 'minimax_h3'
The pack has always shipped apply_gguf_arch_patch.py to fix that, but it was one line buried in the install steps. In practice people hit the error, concluded the models were broken, and gave up — three separate people reported it across the model and workflow repos, and those are just the ones who said something.
KI-generierter Kommentar
KI-generierte Erklärung auf Grundlage von Quellen- und Konfigurationsdetails. Vorschläge sind klar gekennzeichnet.
Use this MiniMax H3 ComfyUI workflow pack to turn a story prompt into a multi-shot video rendered as one continuous take with sound.
The ZIP contains three graphs: H3_Keyframes.json, H3_Multishot_AIO.json, and H3_Multishot_MEMORY.json.
At startup, the pack adds `minimax_h3` to ComfyUI-GGUF's in-memory architecture list for MiniMax-H3 DiT GGUFs, so the manual patch step is not needed.
Optional inputs include an opening photo for shot 1 and a short voice clip to anchor the speaker's voice.
The guide calls for ComfyUI, the MiniMax-H3 Multishot node pack, and a MiniMax-H3 checkpoint; it says the node pack is available through Manager or as this version's ZIP.
The graphs list `minimax_h3_audio_vae_fp32.safetensors`, `minimax_h3_video_vae_fp16.safetensors`, and `MiniMax-H3-fl2va-Q5_1.gguf` among their model files.
H3_Keyframes.json and H3_Multishot_MEMORY.json list `MiniMax-H3-encoder-Q4_K_M.gguf`; H3_Multishot_AIO.json lists `MiniMax-H3-encoder-Q5_K_M.gguf`.
Enter your idea in the prompt box or point the workflow to a prompt file; for hand-written shots, choose `passthrough (raw JSON, skip LLM)` and separate shots with `---` lines.
If an H3 text encoder GGUF has a state-dict or vision mismatch with its `-mmproj` file, use H3 Clip Loader (Any) instead of stock CLIPLoaderGGUF; keep the paired files in the same folder and do not rename them.
Vorschlag · nicht geprüft
Muss das bearbeitet werden?
Melden Sie sich an, um eine Änderungsanfrage zu senden.
Quellen
1 QuelleQuellenauszüge
1 AuszugSource context: 10753 downloads · Type Workflows · Base model MiniMax H3
The pack now does it for you. On startup it adds the architecture to ComfyUI-GGUF's live list in memory. Nothing is written to disk, it is idempotent, and unlike the on-disk patch it survives ComfyUI-GGUF updates instead of being reverted by them. You will see this in the console:
[H3] taught ComfyUI-GGUF the 'minimax_h3' architecture
The old script is still in the folder as a fallback for unusual setups, but you should not need it. If you previously ran it, nothing breaks — the new code sees the architecture is already known and does nothing.
The other GGUF error: text encoder and the mmproj file
Different problem, same week, so it is worth spelling out here. If your H3 text encoder GGUF fails with a state_dict or vision mismatch against its -mmproj file, load it with this pack's H3 Clip Loader (Any) rather than the stock CLIPLoaderGGUF .
The H3 encoder is a truncated Qwen3-VL-32B — 50 layers, no final norm, no lm_head — and its vision tower ships separately as the -mmproj-F16.gguf sidecar. Stock ComfyUI-GGUF only merges an mmproj when the encoder's architecture is qwen2vl . Qwen3-VL reports qwen3vl , so the sidecar is never merged at all, and the missing vision tensors surface as a state_dict mismatch. Its mmproj key map is qwen2vl-era besides: wrong merger keys, and no rules for H3's deepstack mergers or split QKV.
This pack's loader does all three things stock cannot — truncates the text tower, merges the sidecar explicitly, and renames the vision tensors to H3's layout. Keep the -mmproj file in the same folder as the encoder and do not rename either one: they are paired by filename.
And to answer the question that came up directly: a full .safetensors encoder (fp8, int8, NVFP4-AWQ, whatever) works without any of this because it is a complete, pre-shaped model with the vision tower already inside. That is a property of the container, not of the quantization — NVFP4 is not doing anything special.
Nothing else changed
Same nodes, same workflows, same defaults as v1.4. Existing graphs render identically. Upgrading is just overwriting the folder and restarting.
Thanks to the people who took the time to report the error instead of quietly writing the models off. That is the only reason it got fixed.
Hardware and download-size estimates are not available yet.
Anforderungen
minimax_h3_audio_vae_fp32.safetensors
Datei ungeprüftVAE · SAFETENSORS · Unknown
minimax_h3_video_vae_fp16.safetensors
Datei ungeprüftVAE · SAFETENSORS · Unknown
MiniMax-H3-encoder-Q4_K_M.gguf
Nicht aufgelöstText encoder · Unknown
MiniMax-H3-fl2va-Q5_1.gguf
Nicht aufgelöstCheckpoint · Unknown
