Ollama
One command to download and run a language model on your own machine. The easiest honest answer to "how do I do this without a cloud".
- Self-hosted
- Allowed
- No GPU needed
Mistral NeMo 12B
A genuinely Apache-licensed 12B model that fits on an ordinary desktop card, and the base for a lot of small roleplay fine-tunes.
- Open weights
- Text
- Unknown adult content
At a glance
- Made by
- Mistral AI / NVIDIA
- Family
- Mistral
- Licence
- Apache-2.0
- Size
- 12B parameters
- Context
- 128000 tokens
- Graphics memory
- 8 GB minimum, 12 GB recommended
- Adult content
- UnknownWe have not found a clear answer. We would rather say so.
- On Hugging Face
- mistralai/Mistral-Nemo-Instruct-2407
The sweet spot for people without a 24 GB card: quantised it runs on 8 GB, it holds a long conversation, and the licence is a real open-source licence with no acceptable-use appendix.
This combination is why so many of the community's mid-sized roleplay fine-tunes are built on it rather than on Llama.