Most sutiable LLM to run on a 3080 with 10 GB of VRAM and access to 26 GB of system ram…
- User
Most sutiable LLM to run on a 3080 with 10 GB of VRAM and access to 26 GB of system ram for creative writing.
Assistant
For creative writing on a 3080 with 10GB VRAM and 26GB system RAM, I'd recommend Mistral 7B or Llama 3 8B.
These models balance quality writing capabilities with your hardware constraints. They'll run smoothly on your setup while providing strong creative output, coherent storytelling, and good contextual understanding. The 7-8B parameter size fits well within your VRAM budget while leaving room for longer context windows.
If you want to push for slightly better quality, you could try Llama 3 70B with quantization (4-bit or 8-bit), though you might need to manage context length carefully.
Other models, same or similar prompt
Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.