MiniMax H3 needs about 20.9 GB of memory at Q4_K_M (a 18.2 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 16 quants, from 6.7 GB (Q2_K) to 21.4 GB (Q8_0).
MiniMax H3 – 20.1B omni‑modal generative model
MiniMax H3 is a 20.1B parameter model that supports text, images, video, and audio. It can generate video with native stereo audio up to 2K resolution and 15 seconds duration. The model is available in GGUF format for local inference.
Good at
- Multimodal generation
- Video with stereo audio
- High‑resolution output
Limits
- No vision support in text‑only mode
- Large VRAM required for higher quant levels
- License is other
Best for
- Creating short video clips
- Generating audio‑video content
- Multimodal content creation
Who it’s for. Developers and researchers who need a local, multimodal generative model for video and audio synthesis. The model is suitable for applications that require both visual and auditory output in a single pass.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q2_K | 6.7 GB | 8.0 GB | Tight | Get file → | |
| Q2_K | 6.7 GB | 8.0 GB | Too big | Get file → | |
| Q2_K_XL | 8.1 GB | 9.5 GB | Too big | Get file → | |
| Q3_K | 8.7 GB | 10.3 GB | Too big | Get file → | |
| Q3_K | 8.8 GB | 10.3 GB | Too big | Get file → | |
| Q3_K_XL | 9.6 GB | 11.2 GB | Too big | Get file → | |
| Q4_K | 11.4 GB | 13.2 GB | Too big | Get file → | |
| Q4_K | 11.4 GB | 13.3 GB | Too big | Get file → | |
| Q2_K_M | 13.1 GB | 15.2 GB | Too big | Get file → | |
| Q5_0 | 13.9 GB | 16.1 GB | Too big | Get file → | |
| Q5_0 | 13.9 GB | 16.1 GB | Too big | Get file → | |
| Q6_K | 16.6 GB | 19.0 GB | Too big | Get file → | |
| Q6_K | 16.6 GB | 19.1 GB | Too big | Get file → | |
| Q4_K_M | 18.2 GB | 20.9 GB | Too big | Get file → | |
| Q8_0 | 21.4 GB | 24.5 GB | Too big | Get file → | |
| Q8_0 | 21.4 GB | 24.5 GB | Too big | Get file → |
For 8 GB VRAM, use Q2_K or Q2_K_XL. For 16 GB VRAM, Q3_K or Q4_K are suitable. For 24 GB VRAM, Q8_0 provides the highest quality. Memory estimates are approximate.
What GPU do you need for MiniMax H3?
These figures add the file size, an fp16 KV cache at an 8K context window and about 0.5 GB for the runtime. They are estimates, so leave a little headroom, and expect to use less context or a smaller quant if you are right at the limit.
- 4 GB VRAM: no quant fits at an 8K context.
- 6 GB VRAM: no quant fits at an 8K context.
- 8 GB VRAM: the best quant that fits is Q2_K (6.7 GB file, about 8.0 GB in use).
- 12 GB VRAM: the best quant that fits is Q3_K_XL (9.6 GB file, about 11.2 GB in use).
- 16 GB VRAM: the best quant that fits is Q2_K_M (13.1 GB file, about 15.2 GB in use).
- 24 GB VRAM: the best quant that fits is Q4_K_M (18.2 GB file, about 20.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q8_0 (21.4 GB file, about 24.5 GB in use).
Which MiniMax H3 quant should you choose?
Q4_K_M is the usual starting point for MiniMax H3: it balances file size and quality well. Here it is a 18.2 GB file.
The Q8_0 file is 21.4 GB, about 1.2 times the size of Q4_K_M. It keeps the weights closer to the original, which only matters if you have the memory to spare.
If memory is tight, the smallest file is Q2_K at 6.7 GB, with the quality trade-offs that come with fewer bits. Use it only when memory is very tight. Expect a clear drop in quality, which is usually worse on small models and easier to live with on large ones.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about MiniMax H3
How much VRAM does MiniMax H3 need?
About 20.9 GB at Q4_K_M with an 8K context window. That is the 18.2 GB file, plus the KV cache and roughly 0.5 GB of runtime overhead. Smaller quants need less and longer context needs more.
Can I run MiniMax H3 on an 8 GB GPU?
Yes, with Q2_K: the 6.7 GB file needs about 8.0 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of MiniMax H3 GGUF?
It depends on the quant: Q2_K is 6.7 GB, Q2_K is 6.7 GB, Q2_K_XL is 8.1 GB, Q3_K is 8.7 GB, Q3_K is 8.8 GB, Q3_K_XL is 9.6 GB and 10 more.
Where can I download MiniMax H3 GGUF?
The files are on Hugging Face in the ldov/MiniMax-H3-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
What license does MiniMax H3 use?
Hugging Face lists the license as other. Licenses can differ between the quantized files and the original model, so check the repository and the base model before commercial use.
Can I run MiniMax H3 on 8 GB VRAM?
Yes, the Q2_K and Q2_K_XL quantizations use about 8 GB of VRAM at 8k context.
Does MiniMax H3 support text‑only inference?
The model is designed for multimodal input; text‑only usage is not supported by default.
What is the maximum video length MiniMax H3 can generate?
Up to 15 seconds at 24 FPS with native stereo audio.