Midm-2.0-Base-Instruct needs about 9.2 GB of memory at Q4_K_M (a 7.1 GB file) with an 8K context window, so it fits on a 12 GB GPU. This repository has 31 quants, from 3.0 GB (TQ1_0) to 23.1 GB (BF16). Longer context needs more: at 32K tokens the KV cache alone adds about 6.4 GB.
Midm-2.0-Base-Instruct: 11.5B llama architecture with 32K context
Midm-2.0-Base-Instruct is an 11.5B parameter llama model trained for 32,768 token context. It supports tool usage and is released under the MIT license. The model is available in multiple GGUF quantizations from TQ1_0 to Q8_0.
Good at
- Tool integration
- Long context reasoning
- Instruction following
Limits
- No vision support
- Large memory requirement for high‑bit quant
- Requires 32K context for full capability
Best for
- Chatbot with tool calls
- Long‑form content generation
- Instruction‑driven workflows
Who it’s for. Developers building local LLM applications that need a Korean‑centric model with tool support. The model is suitable for users who can allocate sufficient VRAM and RAM to run higher‑bit quantizations.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| TQ1_0 | 3.0 GB | 5.1 GB | Fits | Get file → | |
| IQ1_S | 3.4 GB | 5.6 GB | Fits | Get file → | |
| TQ2_0 | 3.5 GB | 5.6 GB | Fits | Get file → | |
| IQ1_M | 3.7 GB | 5.8 GB | Fits | Get file → | |
| IQ2_XXS | 3.8 GB | 5.9 GB | Fits | Get file → | |
| IQ2_XS | 4.1 GB | 6.2 GB | Fits | Get file → | |
| IQ2_S | 4.2 GB | 6.3 GB | Fits | Get file → | |
| IQ2_M | 4.4 GB | 6.5 GB | Fits | Get file → | |
| Q2_K_S | 4.5 GB | 6.7 GB | Fits | Get file → | |
| Q2_K_M | 4.7 GB | 6.8 GB | Fits | Get file → | |
| IQ3_XXS | 4.9 GB | 7.0 GB | Fits | Get file → | |
| IQ3_XS | 5.1 GB | 7.2 GB | Fits | Get file → | |
| IQ3_M | 5.6 GB | 7.7 GB | Tight | Get file → | |
| IQ3_S | 5.6 GB | 7.7 GB | Tight | Get file → | |
| Q3_K_S | 5.7 GB | 7.8 GB | Tight | Get file → | |
| Q3_K_M | 5.9 GB | 8.0 GB | Tight | Get file → | |
| IQ4_XS | 6.3 GB | 8.4 GB | Too big | Get file → | |
| Q4_0 | 6.5 GB | 8.6 GB | Too big | Get file → | |
| IQ4_NL | 6.7 GB | 8.8 GB | Too big | Get file → | |
| Q4_K_S | 6.9 GB | 9.0 GB | Too big | Get file → | |
| Q4_K_M | 7.1 GB | 9.2 GB | Too big | Get file → | |
| Q4_1 | 7.2 GB | 9.3 GB | Too big | Get file → | |
| Q5_0 | 7.9 GB | 10.1 GB | Too big | Get file → | |
| Q5_K_S | 8.2 GB | 10.3 GB | Too big | Get file → | |
| Q5_K_M | 8.3 GB | 10.4 GB | Too big | Get file → | |
| Q5_1 | 8.7 GB | 10.8 GB | Too big | Get file → | |
| Q6_K_M | 9.5 GB | 11.6 GB | Too big | Get file → | |
| Q8_0 | 12.3 GB | 14.4 GB | Too big | Get file → | |
| BF16 | 16.9 GB | 19.0 GB | Too big | Get file → | |
| F16 | 16.9 GB | 19.0 GB | Too big | Get file → | |
| BF16 | 23.1 GB | 25.2 GB | Too big | Get file → |
For 8 GB VRAM, use TQ1_0 (file size 2.96 GB, memory 5.1 GB at 8 k context). If you have 12 GB VRAM, Q4_0 (6.5 GB, 8.6 GB memory) is a good balance. For 16 GB VRAM, Q5_0 (7.94 GB, 10.1 GB memory) provides higher precision.
What GPU do you need for Midm-2.0-Base-Instruct?
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: the best quant that fits is IQ2_XXS (3.8 GB file, about 5.9 GB in use).
- 8 GB VRAM: the best quant that fits is Q3_K_M (5.9 GB file, about 8.0 GB in use).
- 12 GB VRAM: the best quant that fits is Q6_K_M (9.5 GB file, about 11.6 GB in use).
- 16 GB VRAM: the best quant that fits is Q8_0 (12.3 GB file, about 14.4 GB in use).
- 24 GB VRAM: the best quant that fits is BF16 (16.9 GB file, about 19.0 GB in use).
- 32 GB VRAM: the best quant that fits is BF16 (23.1 GB file, about 25.2 GB in use).
Which Midm-2.0-Base-Instruct quant should you choose?
Q4_K_M is the usual starting point for Midm-2.0-Base-Instruct: it balances file size and quality well. Here it is a 7.1 GB file.
The Q8_0 file is 12.3 GB, about 1.7 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 3.0 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
Source
Frequently asked questions about Midm-2.0-Base-Instruct
How much VRAM does Midm-2.0-Base-Instruct need?
About 9.2 GB at Q4_K_M with an 8K context window. That is the 7.1 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 Midm-2.0-Base-Instruct on an 8 GB GPU?
Yes, with Q3_K_M: the 5.9 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 Midm-2.0-Base-Instruct GGUF?
It depends on the quant: TQ1_0 is 3.0 GB, IQ1_S is 3.4 GB, TQ2_0 is 3.5 GB, IQ1_M is 3.7 GB, IQ2_XXS is 3.8 GB, IQ2_XS is 4.1 GB and 25 more.
What context length does Midm-2.0-Base-Instruct support?
The GGUF header lists a trained context length of 32,768 tokens (32K). The KV cache grows with the context you use: about 1.6 GB at 8K and 6.4 GB at 32K.
Where can I download Midm-2.0-Base-Instruct GGUF?
The files are on Hugging Face in the Mungert/Midm-2.0-Base-Instruct-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 Midm-2.0-Base-Instruct 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 Midm-2.0-Base-Instruct on 8 GB VRAM?
Yes, the TQ1_0 quantization requires about 5.1 GB of VRAM at 8 k context, which fits within 8 GB.
Does the model support vision tasks?
No, the model has no vision capabilities.
What license does the model use?
The model is released under the MIT license.