Kohlibri 40B needs about 25.8 GB of memory at Q4_K_M (a 24.5 GB file) with an 8K context window, so it fits on a 32 GB GPU. This repository has 21 quants, from 0.4 GB to 25.4 GB (Q4_1). Longer context needs more: at 32K tokens the KV cache alone adds about 3.4 GB.
Kohlibri 40B – German‑only large language model
Kohlibri 40B is a 40.3 billion‑parameter model with a 262144‑token context window. It is a mixture of experts and supports reasoning and tool use. The model is released under Apache‑2.0 and is only tested for German.
Good at
- Reasoning in German
- Tool invocation
- Long‑context generation
Limits
- German only, no vision
- Large VRAM required for full precision
- No active parameter count provided
Best for
- German legal document drafting
- German code generation with tool calls
- Long‑form German storytelling
Who it’s for. Developers and researchers needing a large German‑language model with reasoning and tool‑use capabilities. The model is suitable for applications that can run on high‑VRAM GPUs and do not require multilingual support.
Quants & memory
| Quant | File | Est. memory | vs VRAM | Result | |
|---|---|---|---|---|---|
| — | 0.4 GB | 1.8 GB | Fits | Get file → | |
| IQ1_S | 8.4 GB | 9.7 GB | Too big | Get file → | |
| IQ1_M | 9.2 GB | 10.6 GB | Too big | Get file → | |
| IQ2_XXS | 10.7 GB | 12.1 GB | Too big | Get file → | |
| IQ2_XS | 11.9 GB | 13.3 GB | Too big | Get file → | |
| IQ2_S | 12.2 GB | 13.5 GB | Too big | Get file → | |
| IQ2_M | 13.4 GB | 14.7 GB | Too big | Get file → | |
| Q2_K_S | 13.9 GB | 15.2 GB | Too big | Get file → | |
| Q2_K | 14.9 GB | 16.2 GB | Too big | Get file → | |
| IQ3_XXS | 15.6 GB | 17.0 GB | Too big | Get file → | |
| IQ3_XS | 16.6 GB | 18.0 GB | Too big | Get file → | |
| Q3_K_S | 17.6 GB | 18.9 GB | Too big | Get file → | |
| IQ3_S | 17.6 GB | 18.9 GB | Too big | Get file → | |
| IQ3_M | 17.9 GB | 19.2 GB | Too big | Get file → | |
| Q3_K_M | 19.4 GB | 20.8 GB | Too big | Get file → | |
| Q3_K_L | 21.0 GB | 22.4 GB | Too big | Get file → | |
| IQ4_XS | 21.6 GB | 23.0 GB | Too big | Get file → | |
| Q4_0 | 23.0 GB | 24.3 GB | Too big | Get file → | |
| Q4_K_S | 23.1 GB | 24.4 GB | Too big | Get file → | |
| Q4_K_M | 24.5 GB | 25.8 GB | Too big | Get file → | |
| Q4_1 | 25.4 GB | 26.7 GB | Too big | Get file → |
For 8 GB VRAM, use IQ1_S (8.36 GB, 9.7 GB memory at 8k context). For 12 GB VRAM, IQ2_XS (11.95 GB, 13.3 GB memory). For 24 GB VRAM, Q4_K_M (24.5 GB, 25.8 GB memory). These estimates are approximate.
What GPU do you need for Kohlibri 40B?
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 (0.4 GB file, about 1.8 GB in use).
- 6 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.8 GB in use).
- 8 GB VRAM: the best quant that fits is the main file (0.4 GB file, about 1.8 GB in use).
- 12 GB VRAM: the best quant that fits is IQ1_M (9.2 GB file, about 10.6 GB in use).
- 16 GB VRAM: the best quant that fits is Q2_K_S (13.9 GB file, about 15.2 GB in use).
- 24 GB VRAM: the best quant that fits is IQ4_XS (21.6 GB file, about 23.0 GB in use).
- 32 GB VRAM: the best quant that fits is Q4_1 (25.4 GB file, about 26.7 GB in use).
Which Kohlibri 40B quant should you choose?
Q4_K_M is the usual starting point for Kohlibri 40B: it balances file size and quality well. Here it is a 24.5 GB file.
If memory is tight, the smallest file is the first one in the table at 0.4 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 Kohlibri 40B
How much VRAM does Kohlibri 40B need?
About 25.8 GB at Q4_K_M with an 8K context window. That is the 24.5 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 Kohlibri 40B on an 8 GB GPU?
Yes, with the smaller files: the 0.4 GB file needs about 1.8 GB at 8K context. Larger quants in this repository do not fit in 8 GB.
What is the file size of Kohlibri 40B GGUF?
It depends on the quant: file is 0.4 GB, IQ1_S is 8.4 GB, IQ1_M is 9.2 GB, IQ2_XXS is 10.7 GB, IQ2_XS is 11.9 GB, IQ2_S is 12.2 GB and 15 more.
What context length does Kohlibri 40B support?
The GGUF header lists a trained context length of 262,144 tokens (256K). The KV cache grows with the context you use: about 0.8 GB at 8K and 3.4 GB at 32K.
Where can I download Kohlibri 40B GGUF?
The files are on Hugging Face in the mradermacher/Kohlibri-40B-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 Kohlibri 40B 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.
Does the model support English?
No, the model is only tested for German and does not support English.
What license applies to the weights?
The weights are released under Apache‑2.0, the same license as the parent model.