MiniMax-3.1-Flash-DeRisked needs about 259.5 GB of memory at Q4_K_M (a 258.0 GB file) with an 8K context window, which is more than a 48 GB GPU holds. This repository has 9 quants, from 92.1 GB (TQ1_0) to 453.6 GB (Q8_0). Longer context needs more: at 32K tokens the KV cache alone adds about 4.0 GB.
MiniMax-3.1-Flash-DeRisked – 426B parameter model with 1M context window
MiniMax-3.1-Flash-DeRisked is a 426B parameter model in the Minimax family. It supports a 1M token context window and includes vision capabilities. The model is a private pre‑release research artifact.
Good at
- Reasoning
- Tools usage
- Vision
Limits
- Private pre‑release, not fully benchmarked
- Requires 65,536 completion tokens max
- License is other
Best for
- Security research
- False refusal testing
- Red‑team analysis
Who it’s for. Researchers and security practitioners who need a model that can handle large contexts and multimodal inputs while testing refusal behavior. The model is not intended for general public use.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 92.1 GB (6 parts) | 93.6 GB | Too big | Get file → | |
| Q2_K | 155.9 GB (6 parts) | 157.4 GB | Too big | Get file → | |
| Q3_K_M | 203.5 GB (6 parts) | 205.0 GB | Too big | Get file → | |
| IQ4_XS | 229.3 GB (6 parts) | 230.8 GB | Too big | Get file → | |
| IQ4_XS | 229.3 GB (6 parts) | 230.8 GB | Too big | Get file → | |
| MXFP4_MOE | 234.1 GB (6 parts) | 235.6 GB | Too big | Get file → | |
| Q4_K_M | 258.0 GB (6 parts) | 259.5 GB | Too big | Get file → | |
| BF16 | 259.0 GB (6 parts) | 260.5 GB | Too big | Get file → | |
| Q8_0 | 453.6 GB (6 parts) | 455.1 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (file size 258.01 GB, estimated 259.5 GB memory at 8 k context). For 16 GB VRAM, use Q3_K_M (203.49 GB, 205 GB memory). For larger budgets, Q8_0 provides full precision.
What GPU do you need for MiniMax-3.1-Flash-DeRisked?
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: no quant fits at an 8K context.
- 32 GB VRAM: no quant fits at an 8K context.
- 48 GB VRAM: no quant fits at an 8K context.
Which MiniMax-3.1-Flash-DeRisked quant should you choose?
Q4_K_M is the usual starting point for MiniMax-3.1-Flash-DeRisked: it balances file size and quality well. Here it is a 258.0 GB file.
The Q8_0 file is 453.6 GB, about 1.8 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 TQ1_0 at 92.1 GB, with the quality trade-offs that come with fewer bits. Lower bit counts mean smaller files and more quality loss.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Frequently asked questions about MiniMax-3.1-Flash-DeRisked
How much VRAM does MiniMax-3.1-Flash-DeRisked need?
About 259.5 GB at Q4_K_M with an 8K context window. That is the 258.0 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-3.1-Flash-DeRisked on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 93.6 GB, so you would need a larger GPU or CPU offloading.
What is the file size of MiniMax-3.1-Flash-DeRisked GGUF?
It depends on the quant: TQ1_0 is 92.1 GB, Q2_K is 155.9 GB, Q3_K_M is 203.5 GB, IQ4_XS is 229.3 GB, IQ4_XS is 229.3 GB, MXFP4_MOE is 234.1 GB and 3 more.
What context length does MiniMax-3.1-Flash-DeRisked support?
The GGUF header lists a trained context length of 1,048,576 tokens (1024K). The KV cache grows with the context you use: about 1.0 GB at 8K and 4.0 GB at 32K.
Where can I download MiniMax-3.1-Flash-DeRisked GGUF?
The files are on Hugging Face in the freakyskittle/MiniMax-3.1-Flash-DeRisked-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-3.1-Flash-DeRisked 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 this model on 8 GB VRAM?
Use the Q4_K_M quantization; it requires approximately 259.5 GB memory at 8 k context, which is beyond 8 GB VRAM. The model is not feasible on 8 GB VRAM.
Does the model support vision?
Yes, vision is supported via a separate mmproj GGUF file. Image understanding works through llama-mtmd-cli or llama-server.
What is the maximum completion token limit?
The model supports at most 65536 completion tokens, despite upstream claims of 1M context.