MiniMax H3 encoder needs about 22.4 GB of memory at Q4_K_M (a 19.8 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 2 quants, from 19.8 GB (Q4_K_M) to 23.2 GB (Q5_K_M). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
MiniMax H3 encoder – 33B Qwen3-VL text encoder for multimodal tasks
The MiniMax H3 encoder is a 33‑billion‑parameter Qwen3-VL text encoder. It is part of the MiniMax H3 multimodal system and supports coding and vision tasks. The encoder is quantized to Q4_K_M and Q5_K_M for use with llama.cpp.
Good at
- Coding tasks
- Vision‑related text encoding
- Multimodal instruction parsing
Limits
- No vision support without the mmproj sidecar
- Requires H3‑Context‑IR for best results
- License not specified
Best for
- Integrating into ComfyUI pipelines
- Generating video/audio from multimodal prompts
- Fine‑tuning on custom multimodal datasets
Who it’s for. Developers building multimodal applications who need a lightweight text encoder for the MiniMax H3 system. It is suitable for users familiar with llama.cpp and ComfyUI.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q4_K_M | 19.8 GB | 22.4 GB | Too big | Get file → | |
| Q5_K_M | 23.2 GB | 25.9 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (file size 19.76 GB, estimated 22.4 GB memory at 8 k context). For 12 GB VRAM, Q5_K_M (23.21 GB, estimated 25.9 GB memory) is acceptable. Larger VRAM allows the larger quant.
What GPU do you need for MiniMax H3 encoder?
These figures add the file size, an fp16 KV cache at an 8K context window (worked out from this model’s own header) 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: no quant fits at an 8K context.
- 12 GB VRAM: no quant fits at an 8K context.
- 16 GB VRAM: no quant fits at an 8K context.
- 24 GB VRAM: the best quant that fits is Q4_K_M (19.8 GB file, about 22.4 GB in use).
- 32 GB VRAM: the best quant that fits is Q5_K_M (23.2 GB file, about 25.9 GB in use).
Which MiniMax H3 encoder quant should you choose?
Q4_K_M is the usual starting point for MiniMax H3 encoder: it balances file size and quality well. Here it is a 19.8 GB file.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about MiniMax H3 encoder
How much VRAM does MiniMax H3 encoder need?
About 22.4 GB at Q4_K_M with an 8K context window. That is the 19.8 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 encoder on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 22.4 GB, so you would need a larger GPU or CPU offloading.
What is the file size of MiniMax H3 encoder GGUF?
It depends on the quant: Q4_K_M is 19.8 GB, Q5_K_M is 23.2 GB.
What context length does MiniMax H3 encoder support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 2.1 GB at 8K and 8.6 GB at 32K.
Where can I download MiniMax H3 encoder GGUF?
The files are on Hugging Face in the joeygambino/MiniMax-H3-encoder-GGUF repository. The download buttons on this page open the file on Hugging Face. ggufmodels.com does not host any model files.
Can I run the MiniMax H3 encoder on 8 GB VRAM?
Yes, using the Q4_K_M quantization. It requires about 22.4 GB memory at 8 k context, so 8 GB VRAM is insufficient; you need at least 24 GB VRAM for safe operation.
Do I need the mmproj sidecar for vision tasks?
Yes. The encoder’s vision tower is shipped separately as the -mmproj-F16.gguf file and must be kept in the same folder.
Is the model licensed for commercial use?
The license is not specified in the provided data; check the original repository for usage terms.