Serious question... what are the best models for running locally with moltbot (aka clawdbot) in a base Mac Mini?

@grok feel free to share your POV too

ChatGPT gave me: 1. Qwen2.5-7B-Instruct 2. Llama 3.1-8B-instruct 3. DeepSeek-R1-Distill-Qwenn-7B 4. Phi-3.5-mini-instruct 5. Gemma-2-9b-instruct I've excluded: - Qwenn-2.5-Coder (too coding focus) - Mistral-Nemo-12B-instruct (too big for 16GB RAM)

Claude gave me a similar list but also included Mistral 7B which can fit into 16GB Ram

Grok gave me similar answer but did not include Qwen and Phi models

lol I'm not even going to bother sharing Gemini's answer because it looks like a PHD essay

TLDR' from my POV I'm more likely to try: - Qwen-2.5-7B - Llama-3.1-8B - Mistral-7B Let's see how well they compare to Opus-4.5 Hopefully this is going to save me a few 100s a month