MiniMax H3 · Local
Weights are downloadable and ComfyUI has native support, so a MiniMax H3 local install is mostly just files in folders. What catches people is not the diffusion model — it’s the encoder sitting next to it.
Size comparisons almost always quote the diffusion weights alone. But H3's text encoder is Qwen3-VL 32B, and it has to be resident while your prompt is encoded. That single file is frequently larger than the generation model people sized their card against — and it is why a machine that 'should' fit H3 on paper runs out of memory in practice.
If you are tight on memory, the INT8 or NVFP4 encoder builds buy back more headroom than dropping the diffusion model another quant level, because the encoder is the larger of the two.
The official repack is packaged to drop straight into three ComfyUI folders. You do not need every file in the first group — pick the variants that match your workflow and your memory budget.
The generation model itself. Pick ONE fl2va file and/or ONE ref2va file depending on which workflows you plan to run.
minimax_h3_fl2va_bf16.safetensorspick thisFull precision, first/last frame → video + audiominimax_h3_fl2va_int8_convrot.safetensorsINT8, smaller and faster to loadminimax_h3_fl2va_pruned_bf16.safetensorsPruned bf16minimax_h3_fl2va_pruned_int8_convrot.safetensorsPruned INT8 — around 40% smaller than standard INT8minimax_h3_fl2va_pruned_fp8_scaled.safetensorsPruned FP8minimax_h3_ref2va_bf16.safetensorspick thisFull precision, reference-drivenminimax_h3_ref2va_int8_convrot.safetensorsINT8 reference-drivenminimax_h3_ref2va_pruned_bf16.safetensorsPruned bf16 reference-drivenminimax_h3_ref2va_pruned_int8_convrot.safetensorsPruned INT8 reference-drivenminimax_h3_ref2va_pruned_fp8_scaled.safetensorsPruned FP8 reference-drivenRequired. This is a 32B model — see the note below, it is the single biggest reason people underestimate what H3 needs.
qwen3vl_32b_minimax_h3_bf16.safetensorsFull precision, 32B parametersqwen3vl_32b_minimax_h3_int8_convrot.safetensorspick thisINT8 — the practical choice on consumer hardwareqwen3vl_32b_minimax_h3_nvfp4_awq.safetensorsNVFP4 AWQ, smallest of the threeBoth required. H3 generates picture and sound together, so it needs a VAE for each.
minimax_h3_video_vae_fp16.safetensorspick thisVideo VAEminimax_h3_audio_vae_fp32.safetensorspick thisAudio VAESource: Comfy-Org/MiniMax-H3 (official ComfyUI repack) · requires ComfyUI 0.30.0 or later.
The set below is the one ComfyUI's own tutorial specifies — the pruned INT8 diffusion model and the NVFP4 text encoder, not the bf16 files. It is the smallest official combination that runs all three workflows minus reference-to-video.
Add the ref2va diffusion model on top if you want reference-to-video — it is a second checkpoint of comparable size, not a small add-on.
Source: File list per docs.comfy.org · sizes per github.com/Tomiigo/minimax-h3-16gb
A local install is worth it if you generate constantly and want full control. H3Video’s hosted text-to-video generator provides a lower-setup option when you do not want to download or maintain the local model stack.
Compare hosted pricing →Once the files are in place, the ComfyUI workflow guide covers the three templates and the errors that show up first. If memory is your constraint, the VRAM breakdown explains what actually allocates during generation, and the MiniMax H3 LoRA guide can cut your sampling time by roughly 5×, and it works with whichever file you picked above — the publisher supports the pruned and curve variants as well as the full ones.
With the install done, the thing that decides whether your first clip is any good is the prompt, not the checkpoint. The MiniMax H3 prompt guide has the three-field format the model was trained on and a library of prompts written in it — including the audio block that stops H3 inventing a soundtrack for you.
Four things: one diffusion model (fl2va for frame-driven, ref2va for reference-driven), the Qwen3-VL 32B text encoder, the video VAE and the audio VAE. The encoder and both VAEs are non-optional — the diffusion model alone will not run.
Diffusion models go in ComfyUI/models/diffusion_models/, the text encoder in ComfyUI/models/text_encoders/, and both VAE files in ComfyUI/models/vae/. The Comfy-Org repack is packaged specifically to drop into those three folders.
Because the text encoder is Qwen3-VL 32B. Most size comparisons quote only the diffusion weights, but the encoder must be resident while your prompt is encoded and is frequently the larger of the two files. That is why machines that look sufficient on paper still run out of memory.
Pruned INT8 checkpoints are roughly 40% smaller than standard INT8 thanks to precomputed adaLN curve tables. They are a good way to save space, and they do not lock you out of the turbo LoRA — the publisher supports pruned and curve variants alongside the full ones, with the ComfyUI node detecting a pruned base automatically.