en.Wedoany.com Reported - Open model runtime tool Ollama has completed a $65 million Series B funding round, led by Theory Ventures, with participation from Benchmark, 8VC, Y Combinator, and others. The company has raised a total of $88 million since its founding three years ago.
Ollama's philosophy is to allow developers to download and run open-weight models locally with a single command. If a local device cannot handle larger models, its cloud service will run them with the same setup as a substitute, charging based on GPU time rather than per token. Founder and CEO Jeff Morgan shared details with TechCrunch, which first reported on this funding round.
The founding team has similar experience. Morgan and co-founder Michael Chiang previously developed Kitematic, which was acquired by Docker in 2015. Their work evolved into Docker Desktop, which now serves over ten million developers. Ollama hides the cumbersome setup steps in the AI field, enabling users to run models directly. The company reports that it currently has 8.9 million monthly active developers, up from 4.5 million in January. Weekly new installations are approaching one million. Its product has been adopted by 85% of Fortune 500 companies, covering government, healthcare, and financial sectors. The entire team consists of just 14 people.
This funding round bets on the shift from closed AI models to open models. Benchmark partner Peter Fenton stated that open-weight models will generate the vast majority of tokens over the next 18 to 24 months. Tomasz Tunguz of Theory Ventures views Ollama as a platform layer into which other applications can plug. Fenton believes it's not a binary choice between open and closed, but enterprises with high inference costs have strong incentives to move to open models and can rely on closed models when needed. Jeff Morgan noted that the turning point came around January, when open models became good enough to handle agentic tasks like coding.
Ollama's cloud platform hosts heavyweight open models such as Nemotron, GLM, DeepSeek, Kimi, and MiniMax. It has distribution partnerships with these labs as well as chip manufacturers Nvidia, AMD, Intel, and Qualcomm, allowing users to access new models on day one.
Ollama's growth has also drawn criticism. About a year ago, some users accused it of diverting attention from free projects to paid cloud services, calling it a "degradation" of developer tools. Morgan and Fenton countered that the free desktop application has not changed, and the cloud service is only used to run larger models that cannot be handled locally.
