Mellum2.1 12B A2.5B Thinking needs about 9.8 GB of memory at Q4_K_M (a 8.8 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 23 quants, from 0.1 GB to 24.3 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 1.9 GB.
Mellum2.1 12B A2.5B Thinking – a 12B mixture‑of‑experts model focused on reasoning and agentic tasks
Mellum2.1 12B A2.5B Thinking is a 12 billion‑parameter model with 2.5 billion active parameters. It supports a 131 072‑token context and is trained for complex reasoning, coding, math, and tool‑using tasks. The model is released under Apache 2.0.
Good at
- Advanced reasoning and problem solving
- Agentic code editing and tool use
- Long‑form context handling
Limits
- Requires large VRAM for higher‑quality quants
- No vision or audio input
- Licensed under Apache 2.0 – no commercial restrictions
Best for
- Developers building autonomous coding assistants
- Researchers testing long‑context reasoning
- Teams needing a large‑context LLM for codebase analysis
Who it’s for. The model is suitable for developers, researchers, and teams that need a powerful reasoning engine for text‑only tasks. It is ideal for building agents that interact with repositories, run commands, and solve complex problems.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.1 GB | 1.1 GB | Fits | Get file → | |
| IQ2_M | 5.4 GB | 6.4 GB | Fits | Get file → | |
| Q2_K | 5.5 GB | 6.5 GB | Fits | Get file → | |
| Q3_K_S | 6.1 GB | 7.0 GB | Fits | Get file → | |
| IQ3_XXS | 6.3 GB | 7.3 GB | Tight | Get file → | |
| IQ3_XS | 6.3 GB | 7.3 GB | Tight | Get file → | |
| Q3_K_M | 6.7 GB | 7.7 GB | Tight | Get file → | |
| IQ4_XS | 7.0 GB | 7.9 GB | Tight | Get file → | |
| Q4_0 | 7.0 GB | 8.0 GB | Tight | Get file → | |
| Q3_K_L | 7.1 GB | 8.1 GB | Too big | Get file → | |
| IQ3_M | 7.2 GB | 8.2 GB | Too big | Get file → | |
| IQ4_NL | 7.7 GB | 8.6 GB | Too big | Get file → | |
| Q4_1 | 7.7 GB | 8.7 GB | Too big | Get file → | |
| Q4_K_S | 7.9 GB | 8.8 GB | Too big | Get file → | |
| Q4_K_M | 8.8 GB | 9.8 GB | Too big | Get file → | |
| Q5_K_S | 9.0 GB | 9.9 GB | Too big | Get file → | |
| Q4_K_L | 9.2 GB | 10.2 GB | Too big | Get file → | |
| Q5_K_M | 10.0 GB | 10.9 GB | Too big | Get file → | |
| Q6_K_S | 11.1 GB | 12.1 GB | Too big | Get file → | |
| Q6_K | 11.5 GB | 12.4 GB | Too big | Get file → | |
| Q6_K_L | 11.8 GB | 12.8 GB | Too big | Get file → | |
| Q8_0 | 12.9 GB | 13.9 GB | Too big | Get file → | |
| BF16 | 24.3 GB | 25.3 GB | Too big | Get file → |
For 8 GB VRAM, use Q4_K_M (8.82 GB) or Q4_0 (7 GB) for a balance of size and performance. For 12 GB VRAM, Q6_K (11.45 GB) or Q6_K_S (11.13 GB) provide higher precision. For 16 GB VRAM, Q8_0 (12.93 GB) offers the best quality. Memory estimates are approximate.
What GPU do you need for Mellum2.1 12B A2.5B Thinking?
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: the best quant that fits is the main file (0.1 GB file, about 1.1 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.1 GB file, about 1.1 GB in use).
- 8 GB VRAM: the best quant that fits is Q4_0 (7.0 GB file, about 8.0 GB in use).
- 12 GB VRAM: the best quant that fits is Q5_K_M (10.0 GB file, about 10.9 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (12.9 GB file, about 13.9 GB in use).
- 24 GB VRAM: the best quant that fits is Q8_0 (12.9 GB file, about 13.9 GB in use).
- 32 GB VRAM: the best quant that fits is BF16 (24.3 GB file, about 25.3 GB in use).
Which Mellum2.1 12B A2.5B Thinking quant should you choose?
Q4_K_M is the usual starting point for Mellum2.1 12B A2.5B Thinking: it balances file size and quality well. Here it is a 8.8 GB file.
The Q8_0 file is 12.9 GB, about 1.5 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 the first one in the table at 0.1 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 Mellum2.1 12B A2.5B Thinking
How much VRAM does Mellum2.1 12B A2.5B Thinking need?
About 9.8 GB at Q4_K_M with an 8K context window. That is the 8.8 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 Mellum2.1 12B A2.5B Thinking on an 8 GB GPU?
Yes, with Q4_0: the 7.0 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 Mellum2.1 12B A2.5B Thinking GGUF?
It depends on the quant: file is 0.1 GB, IQ2_M is 5.4 GB, Q2_K is 5.5 GB, Q3_K_S is 6.1 GB, IQ3_XXS is 6.3 GB, IQ3_XS is 6.3 GB and 17 more.
What context length does Mellum2.1 12B A2.5B Thinking support?
The GGUF header lists a trained context length of 131,072 tokens (128K). The KV cache grows with the context you use: about 0.5 GB at 8K and 1.9 GB at 32K.
Where can I download Mellum2.1 12B A2.5B Thinking GGUF?
The files are on Hugging Face in the bartowski/Mellum2.1-12B-A2.5B-Thinking-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 Mellum2.1 12B A2.5B Thinking 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 Mellum2.1 on 8 GB VRAM?
Yes, use Q4_K_M or Q4_0 quants, which fit within 8 GB VRAM and provide reasonable performance.
Does Mellum2.1 support image input?
No, the model only supports text input and does not have vision capabilities.
What license applies to Mellum2.1?
Mellum2.1 is released under the Apache 2.0 license, allowing free use and modification.