Local models, developer control, and the future of AI runtimes | BRK235
Ollama introduces hybrid local-cloud inference and a 'launch' command to run open models in agent tools.
“Ola is really the easiest way for developers to access open models and use it with your own tools.”
At Microsoft Build (BRK235), Ollama co-founder Michael and Parth detailed how open models power agents via hybrid local-cloud inference, a new 'launch' command that injects open models into tools like VS Code and GitHub Copilot, and Ollama Cloud with zero data retention. They cited over 8 million active developers and noted Google DeepMind's Gemma 4 12B is already available on the platform. It matters as a signal that open-model tooling is maturing into agent-ready infrastructure with frontier-grade cloud scaling.