Ollama raises $65 million to bring AI models to PCs

Ollama raised $65 million in a Series B led by Theory Ventures, lifting total funding to $88 million as more developers move AI workloads off cloud platforms and onto their own PCs. The San Francisco company, launched in 2023, reaches more than 8.9 million developers every month and sits inside 85% of the Fortune 500, while the business itself still runs with just 14 employees.

Ollama makes open-weight models easy to run locally in minutes, then offers cloud access when developers need larger models or heavier workloads. The local-first model offers lower costs, more control over data and fewer setup headaches. Ollama’s GitHub footprint reflects that momentum, with 176,000 stars and nearly 17,000 forks.

Ollama’s cloud service adds a separate revenue path through free, Pro and Max tiers priced at $20 a month or $200 a year, and $100 a month. The company’s cloud models run in the United States, Europe and Singapore, and customer data is not trained on cloud models. The privacy promise, along with the option to keep work local, is central to the company’s challenge to closed AI ecosystems built around centralized compute.

Jeff Morgan and co-founder Michael Chiang came to the company with experience building developer tools, including Docker Desktop after Docker acquired their earlier startup, Kitematic. Ollama’s “Docker for AI” framing abstracts away hardware and setup friction so developers can treat open models more like software infrastructure than research projects. Morgan said the business inflection point came around January 2026, when larger open models became good enough for agentic coding tasks and the market started shifting toward open models for real work, especially coding and enterprise use cases.

The product line has expanded alongside the user base. Ollama launched a new macOS and Windows app on July 30, 2025, with file chat and multimodal support. It previewed a faster Apple Silicon version powered by MLX on March 30, 2026, and it introduced a command on January 23, 2026 for running coding tools such as Claude Code, OpenCode and Codex with local or cloud models. Its documentation now includes local and cloud models, model browsing, integrations, and Python and JavaScript libraries.