The Best AI Models to Run Locally (According to My Totally Biased Opinion)
The Best AI Models to Run Locally (According to My Totally Biased Opinion)
I’ve spent a lot of time running local AI models, and after extensive testing (read: lots of random chats and questionable prompts), I’ve settled on my favorites. No benchmarks, no pointless metrics—just pure vibes and personal preference.
🏆 Best Overall: Llama 3.1 8B
If you want a general-purpose AI that does everything well, Llama 3.1 8B is the clear winner. It’s smart, coherent, and doesn’t go off the rails too easily. Whether you’re chatting, building a assistant, or writing this model just works.
🔬 Best Technical Model: DeepSeek-R1 8B
Need something with a more to-the-point attitude? DeepSeek-R1 8B is the one. It’s less verbose, highly technical, and great for when you need precise answers instead of a paragraph-long essay. Perfect for hacking, cybersecurity, and other deep technical topics.
⚡ Best for Speed (and Dumb Fun): Qwen 1.5 0.5B
If you don’t need the smartest model and just want something fast, Qwen 1.5 0.5B is a solid choice. It responds quickly, makes decent small talk, and is good for simple chatbot-style interactions. Not the best at deep reasoning, but hey, sometimes you just need a dumb bot that replies fast.
💻 Best for Coding: Starcoder2 7B
Despite newer models dropping left and right, Starcoder2 7B is still the local coding model. It understands context well, generates clean code, and doesn’t hallucinate random syntax as much as other models. If you’re working on a coding project, this is still the GOAT.
These are my go-to local AI models. If you disagree, that’s fine—just know that you’re wrong. Try them out yourself and let me know if you find any hidden gems worth running.
Happy local AI tinkering! 🚀
Advocating for privacy does not finance itself. If you enjoyed this article, please consider zapping or sending monero
82XCDNK1Js8TethhpGLFPbVyKe25DxMUePad1rUn9z7V6QdCzxHEE7varvVh1VUidUhHVSA4atNU2BTpSNJLC1BqSvDajw1