Runitlocal.io – Run AI on your own hardware

Your Mac can run AI right now.

Qwen3 4B at about 22 tok/s, about reading speed. Quick everyday chat and summaries.

First time? Install Ollama, then paste that into a terminal.

Safari doesn't say which chip your Mac has. Pick it for a sharper answer.

Everything your Mac can run

With Ollama, at its default 4-bit size. Speeds are estimates from memory bandwidth, usually within about a third of real results.

• Qwen3 4BQuick everyday chat and summaries~22 tok/sUsableqwen3:4b

• Qwen3 8BA solid all-rounder for chat and writing~11 tok/sUsableqwen3:8b

• Gemma 3 12BReads images as well as text~7 tok/sSlowgemma3:12b

• Qwen3 14BNoticeably smarter at reasoning and code~6 tok/sSlowqwen3:14b

• gpt-oss 20BOpenAI's open model. Fast, strong reasoningneeds 16 GBToo biggpt-oss:20b

• Gemma 3 27BGoogle's best open model for its sizeneeds 19 GBToo biggemma3:27b

• Qwen3 30B-A3B30B smarts at small-model speedneeds 21 GBToo bigqwen3:30b

• Qwen3 32BStrong coding and long answersneeds 22 GBToo bigqwen3:32b

• Llama 3.3 70BClose to the big cloud modelsneeds 45 GBToo bigllama3.3:70b

• gpt-oss 120BOpenAI's largest open modelneeds 67 GBToo biggpt-oss:120b

• Qwen3 235B-A22BAlibaba's flagship open model, for multi-GPU rigsneeds 144 GBToo bigqwen3:235b

• DeepSeek R1 671BThe full DeepSeek R1, for the biggest rigsneeds 406 GBToo bigdeepseek-r1:671b

Your GPU is read by your browser and never sent to us. Speeds are estimates unless marked measured. Measured something on your own rig? Send it to us.

Everything else, one click away