Mammoth 1 needs about 19.2 GB of memory at Q4_K_M (a 16.5 GB file) with an 8K context window, so it fits on a 24 GB GPU. This repository has 5 quants, from 14.3 GB (Q3_K_L) to 53.8 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 8.6 GB.
Mammoth 1 – 26.9B parameter decision model with vision, reasoning, and tools
Mammoth 1 is a 26.9B parameter decision model based on Qwen3.5. It supports text and image inputs and returns calibrated probabilities for multiple-choice, yes/no, and ordered level questions. The model uses a block-bidirectional attention mask and a shared-prefix prefill up to 16,384 tokens.
Good at
- Decision making with calibrated probabilities
- Multi-modal input (text + images)
- Single-pass multi-question evaluation
Limits
- No vision support beyond images
- Requires CUDA GPU for self-hosting
- License is MIT – no restrictions on commercial use
Best for
- Building routing or guardrail systems
- Deploying decision engines in production
- Evaluating multiple-choice questions in a single pass
Who it’s for. Developers building decision or routing systems that need calibrated probabilities. Researchers needing a high-performance decision model with vision support. Teams that require a single-pass evaluation of multiple questions.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| Q3_K_L | 14.3 GB | 17.0 GB | Too big | Get file → | |
| Q4_K_M | 16.5 GB | 19.2 GB | Too big | Get file → | |
| Q5_K_M | 19.2 GB | 21.9 GB | Too big | Get file → | |
| Q6_K | 22.1 GB | 24.7 GB | Too big | Get file → | |
| BF16 | 53.8 GB | 56.5 GB | Too big | Get file → |
For 8 GB VRAM use Q3_K_L (14.34 GB, 17 GB memory at 8k context). For 12 GB VRAM use Q4_K_M (16.55 GB, 19.2 GB memory). For 16 GB VRAM use Q5_K_M (19.23 GB, 21.9 GB memory).
What GPU do you need for Mammoth 1?
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 Q5_K_M (19.2 GB file, about 21.9 GB in use).
- 32 GB VRAM: the best quant that fits is Q6_K (22.1 GB file, about 24.7 GB in use).
- 48 GB VRAM: the best quant that fits is Q6_K (22.1 GB file, about 24.7 GB in use).
Which Mammoth 1 quant should you choose?
Q4_K_M is the usual starting point for Mammoth 1: it balances file size and quality well. Here it is a 16.5 GB file.
If memory is tight, the smallest file is Q3_K_L at 14.3 GB, with the quality trade-offs that come with fewer bits. A reasonable choice when a 4-bit file will not fit, mostly on larger models. Quality loss is noticeable but often acceptable.
New to quant names? Read GGUF quantization explained.
Specifications
Architecture from the GGUF header
Source
Frequently asked questions about Mammoth 1
How much VRAM does Mammoth 1 need?
About 19.2 GB at Q4_K_M with an 8K context window. That is the 16.5 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 Mammoth 1 on an 8 GB GPU?
Not at an 8K context. The smallest file needs about 17.0 GB, so you would need a larger GPU or CPU offloading.
What is the file size of Mammoth 1 GGUF?
It depends on the quant: Q3_K_L is 14.3 GB, Q4_K_M is 16.5 GB, Q5_K_M is 19.2 GB, Q6_K is 22.1 GB, BF16 is 53.8 GB.
What context length does Mammoth 1 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 Mammoth 1 GGUF?
The files are on Hugging Face in the prithivMLmods/mammoth-1-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 Mammoth 1 use?
Hugging Face lists the license as mit. 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 Mammoth 1 on 8 GB VRAM?
Yes, using the Q3_K_L quantization file which is 14.34 GB and requires approximately 17 GB of VRAM at 8k context.
Does Mammoth 1 support vision?
Yes, it accepts text and image inputs but the vision encoder is frozen.
What is the maximum context length?
The model supports up to 16,384 tokens in a single pass.