Local Ai
You can now try offline computer use with cua-driver + Muse Glimmer, through our friends at @ollama š¦ How-to: https://cua.ai/docs/how-to-guā¦
Francesco @francedot reported that the first fully offline computerābased LLM experience was achieved by running Muse Glimmer 30B (from @AIatMeta) locally on macOS, using Cua Driver to control native
Francesco @francedot reported that the first fully offline computerābased LLM experience was achieved by running MuseāÆGlimmer 30B (from @AIatMeta) locally on macOS, using CuaāÆDriver to control native apps such as Notes and Reminders. The demonstrationādocumented at https://cua.ai/docs/how-to-guā¦āshowed that a 30ābillionāparameter model can operate without internet access, a milestone previously considered unlikely for such large models.
Related
- Run @NousResearch's Hermes Agent fully locally on DGX Spark. š Our newest playbook shows you how to get set up via @Ollama step by step. š
- https://ollama.com/blog/all-aboard-open-models
- https://docs.ollama.com/integrations/hermes#configure-desktop-app-with-ollama-cloud-models
- Using Ollama Gemma4 models via OpenWebUI on my phone and itās been a good experience
Source: Ollama (X) | 2026-08-10