Marco Mini Base needs about 12.0 GB of memory at Q4_K_M (a 10.6 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 25 quants, from 0.1 GB to 14.3 GB (Q6_K). Longer context needs more: at 32K tokens the KV cache alone adds about 3.8 GB.
17.4B parameter MoE language model with 0.86B active per token
Marco Mini Base is a compact, highly sparse Mixture-of-Experts multilingual language model. It activates only 0.86B out of 17.4B total parameters per token, matching dense models with up to 4B parameters on English and multilingual benchmarks.
Good at
- Sparse MoE inference
- Multilingual support across 29 languages
- Low activation ratio 5%
Limits
- Requires 8k context for memory estimates
- No vision support
- Apache 2.0 license
Best for
- Edge inference on limited GPU
- Multilingual chatbot
- Instruction following
Who it’s for. Developers and researchers who need an efficient multilingual LLM with a low memory footprint. The model is suitable for deployment on systems with limited GPU resources while maintaining strong performance across many languages.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.1 GB | 1.5 GB | Fits | Get file → | |
| IQ1_S | 3.6 GB | 5.1 GB | Fits | Get file → | |
| IQ1_M | 4.0 GB | 5.4 GB | Fits | Get file → | |
| IQ2_XXS | 4.7 GB | 6.1 GB | Fits | Get file → | |
| IQ2_XS | 5.2 GB | 6.6 GB | Fits | Get file → | |
| IQ2_S | 5.2 GB | 6.7 GB | Fits | Get file → | |
| IQ2_M | 5.8 GB | 7.2 GB | Tight | Get file → | |
| Q2_K_S | 6.0 GB | 7.4 GB | Tight | Get file → | |
| Q2_K | 6.4 GB | 7.8 GB | Tight | Get file → | |
| IQ3_XXS | 6.7 GB | 8.2 GB | Too big | Get file → | |
| IQ3_XS | 7.2 GB | 8.6 GB | Too big | Get file → | |
| IQ3_S | 7.6 GB | 9.0 GB | Too big | Get file → | |
| Q3_K_S | 7.6 GB | 9.0 GB | Too big | Get file → | |
| IQ3_M | 7.7 GB | 9.1 GB | Too big | Get file → | |
| Q3_K_M | 8.4 GB | 9.8 GB | Too big | Get file → | |
| Q3_K_L | 9.0 GB | 10.5 GB | Too big | Get file → | |
| IQ4_XS | 9.3 GB | 10.8 GB | Too big | Get file → | |
| IQ4_NL | 9.9 GB | 11.3 GB | Too big | Get file → | |
| Q4_0 | 9.9 GB | 11.3 GB | Too big | Get file → | |
| Q4_K_S | 9.9 GB | 11.4 GB | Too big | Get file → | |
| Q4_K_M | 10.6 GB | 12.0 GB | Too big | Get file → | |
| Q4_1 | 10.9 GB | 12.4 GB | Too big | Get file → | |
| Q5_K_S | 12.0 GB | 13.5 GB | Too big | Get file → | |
| Q5_K_M | 12.4 GB | 13.8 GB | Too big | Get file → | |
| Q6_K | 14.3 GB | 15.7 GB | Too big | Get file → |
For 8 GB VRAM, the "unknown" quant at 0.08 GB file size uses approximately 1.5 GB at 8k context. If more memory is available, the IQ1_S quant (3.63 GB, 5.1 GB memory) or IQ1_M (4.01 GB, 5.4 GB memory) provide a balance between size and performance. For higher quality, consider Q3_K_M (8.36 GB, 9.8 GB memory) if the GPU can accommodate it.
What GPU do you need for Marco Mini Base?
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 1.5 GB in use).
- 6 GB VRAM: the best quant that fits is IQ1_M (4.0 GB file, about 5.4 GB in use).
- 8 GB VRAM: the best quant that fits is Q2_K (6.4 GB file, about 7.8 GB in use).
- 12 GB VRAM: the best quant that fits is Q4_K_M (10.6 GB file, about 12.0 GB in use).
- 16 GB VRAM: the best quant that fits is Q6_K (14.3 GB file, about 15.7 GB in use).
Which Marco Mini Base quant should you choose?
Q4_K_M is the usual starting point for Marco Mini Base: it balances file size and quality well. Here it is a 10.6 GB file.
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 Marco Mini Base
How much VRAM does Marco Mini Base need?
About 12.0 GB at Q4_K_M with an 8K context window. That is the 10.6 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 Marco Mini Base on an 8 GB GPU?
Yes, with Q2_K: the 6.4 GB file needs about 7.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Marco Mini Base GGUF?
It depends on the quant: file is 0.1 GB, IQ1_S is 3.6 GB, IQ1_M is 4.0 GB, IQ2_XXS is 4.7 GB, IQ2_XS is 5.2 GB, IQ2_S is 5.2 GB and 19 more.
What context length does Marco Mini Base support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 0.9 GB at 8K and 3.8 GB at 32K.
Where can I download Marco Mini Base GGUF?
The files are on Hugging Face in the mradermacher/Marco-Mini-Base-i1-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 Marco Mini Base 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 Marco Mini Base on 8 GB VRAM?
The smallest quant (unknown) uses about 1.5 GB at 8k context, while larger quants require more memory. Choose a quant that fits your VRAM budget.
Does Marco Mini Base support vision?
No, the model does not include vision capabilities.
What license is Marco Mini Base released under?
It is released under the Apache 2.0 License.