Skip to content
TheChatArchiveThe Chat Archive
Claudeclaude-3-7-sonnet-20250219-thinking-32k· AnthropicHugging Face LMArena Human Preference 140K

You are an ai expert and you have high knowledge of running ai models locally. Your task…

From a Chatbot Arena battleThe same person sent this prompt to two anonymous models at once. Compare it with Claudeclaude-3-5-sonnet-20241022’s answer.
  1. User

    You are an ai expert and you have high knowledge of running ai models locally. Your task is to advise me for my local setup. I use hardware with a nvidia rtx 4000 ada 20GB RAM. Which model is the best for coding in my given setup? The licence has to allow me to use the model private and commercial.

  2. Claude Assistant

    For your NVIDIA RTX 4000 Ada with 20GB VRAM, I'd recommend CodeLlama-7B or DeepSeek Coder 6.7B as your best options for local coding assistance. Both run well on your hardware when quantized to 4-bit precision and offer excellent coding capabilities.

    CodeLlama-7B is particularly strong for general programming tasks, while DeepSeek Coder has shown impressive benchmark results specifically for code completion and generation.

    If you want something that runs even more efficiently, consider CodeGemma 7B or Phi-2 (only 2.7B parameters but surprisingly capable).

    All these models have permissive licenses allowing both private and commercial use. You can run them using Ollama, llama.cpp, or text-generation-webui for a smooth experience.

    Would you like more details about any of these models or how to set them up locally?

Responses were produced in different, uncontrolled contexts (system prompts, settings, prior turns). This is not a benchmark.