MiniMax H3 TextEncoder 16GB needs about 15.9 GB of memory at its main file (a 13.7 GB file) with an 8K context window, so it fits on a 16 GB GPU. This repository has 3 quants, from 12.6 GB to 14.8 GB. Longer context needs more: at 32K tokens the KV cache alone adds about 6.7 GB.
MiniMax H3 TextEncoder 16GB – text encoder for H3 vision models
MiniMax H3 TextEncoder 16GB is a 25.8B parameter text encoder derived from Qwen3-VL-32B-Instruct. It contains no language model head and is intended for use with the H3 visual tower in ComfyUI. The model supports coding, tools, and vision tasks.
Good at
- Vision‑text encoding
- Coding assistance
- Tool invocation
Limits
- No text generation
- Requires 16 GB GPU
- No CPU fallback
Best for
- ComfyUI H3 pipelines
- Visual tower integration
- Text conditioning for DiT
Who it’s for. Developers building H3 vision‑language pipelines who need a lightweight text encoder that fits on 16 GB GPUs. The model is not suitable for general chat or large‑scale inference on smaller GPUs.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 12.6 GB | 14.8 GB | Too big | Get file → | |
| — | 13.7 GB | 15.9 GB | Too big | Get file → | |
| — | 14.8 GB | 17.0 GB | Too big | Get file → |
For 16 GB GPUs use the 11.5 GB file (file_size_gb 12.65, est_memory_gb_at_8k_context 14.8). A 12.5 GB file (13.7 GB, 15.9 GB memory) also fits. Larger files exceed 16 GB VRAM and will OOM.
What GPU do you need for MiniMax H3 TextEncoder 16GB?
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: the best quant that fits is the main file (13.7 GB file, about 15.9 GB in use).
- 24 GB VRAM: the best quant that fits is the main file (14.8 GB file, about 17.0 GB in use).
Which MiniMax H3 TextEncoder 16GB quant should you choose?
If memory is tight, the smallest file is the first one in the table at 12.6 GB, with the quality trade-offs that come with fewer bits.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about MiniMax H3 TextEncoder 16GB
How much VRAM does MiniMax H3 TextEncoder 16GB need?
About 15.9 GB at its main file with an 8K context window. That is the 13.7 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 TextEncoder 16GB on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 14.8 GB, so you would need a larger GPU or CPU offloading.
What is the file size of MiniMax H3 TextEncoder 16GB GGUF?
It depends on the quant: file is 12.6 GB, file is 13.7 GB, file is 14.8 GB.
What context length does MiniMax H3 TextEncoder 16GB support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 1.7 GB at 8K and 6.7 GB at 32K.
Where can I download MiniMax H3 TextEncoder 16GB GGUF?
The files are on Hugging Face in the pottokao/MiniMax-H3-TextEncoder-16GB-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 TextEncoder 16GB use?
Hugging Face lists the license as apache-2.0. 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 TextEncoder on an 8 GB GPU?
No. The smallest file requires 12.65 GB and estimates 14.8 GB VRAM at 8K context, exceeding 8 GB limits.
Does the model support text generation?
No. It contains no language model head and cannot chat or generate text.
What license applies to this model?
The model is released under the apache-2.0 license.