Ollama Inc., the largest artificial intelligence platform connecting developers to open models, today announced it has raised $65 million in a new funding round led by Theory Ventures.

Benchmark, 8VC, Y Combinator, Pace Capital, 49 Palms, GTMFund, and other investors and angels also participated in the Series B round. Today’s funding brings the company’s total raised to $88 million.

Ollama provides an easy-to-use tool that AI developers and engineers can run locally. With a single command, they can download and deploy AI models on their own local hardware or execute on more powerful models in Ollama’s cloud.

“Open models should be easy to run, easy to build with and available wherever people need them — on your own machine, in the cloud, or both,” said founder and Chief Executive Jeffrey Morgan. “Ollama started as an open-source project, and has since grown into a community of millions of developers.”

Since its launch in 2023, the tool has been used by more than 8.9 million developers each month, making it the largest network in the open model ecosystem and featuring more than 67,000 integrations. The company said it’s used by 85% of Fortune 500 enterprise companies, including customers in deeply regulated industries, such as government, healthcare and finance.

On its front end, Ollama provides a graphical user interface and a command line that make it simple to activate and test models, both locally and in the cloud.

Although the tool initially launched with the capability to provide local AI models and application protocol interface connectors for cloud model providers, the pivot to providing neocloud inference enabled the company to build it natively into the product. Users can sign up for a free tier that offers time on graphics processing units, with subscriptions up to $100 a month.

The company’s cloud inference pricing model competes with other developer-centric AI cloud inference providers such as Together Computer Inc., Fireworks AI Inc. and Groq Inc. that provide access to large models on a per-token pricing basis.

Ollama differentiates itself by offering a cloud service that extends the tool for developers who already run models locally. The same command, the same OpenAI-compatible API that addresses any model provider or local 7B model can also swap to a 400B cloud model with a single string. This allows developers a simple continuity of workflow rather than any separate cloud relationship.

Through a network of partnerships, the company can provide developers release-day access to powerful new open models for experimentation, testing and building. Open models available in the company’s catalog include GLM, Nemotron, DeepSeek, Kimi and MiniMax.

The company said that, on average, the cloud has more than doubled in token volume every month.

The difficulty is that GPU-time metering is difficult to forecast, whereas per-token is easier to control. Specialists such as Fireworks also provide extremely high-speed, low-latency access, and large-catalog providers such as Groq and Together offer high availability of frontier-class models.

Direct competitors to Ollama include LM Studio, which is a polished GUI for browsing models, chatting and visually tuning models. Many developers use Ollama as a nuts-and-bolts, under-the-hood runtime for scripting, servers and app building, while LM Studio is a go-to for GUI-first exploration and experimentation. Other examples include llama.cpp itself, the low-level engine Ollama itself builds on; vLLM, a production-scale multi-GPU tool; Jan, a fully open-source ChatGPT-style desktop app for cloud AI models; and MLX, a purely Apple silicon AI tool.

Many developers use Ollama as a backend for apps, or to power other interfaces such as AnythingLLM, a per-project retrieval augmented generation workspace setup that powers a whole chat and agent client; Open WebUI, a popular Ollama client that self-hosts a ChatGPT-like UI with RAG, voice and plugins; and full-on local AI agents such as OpenClaw and Hermes.

Image: SiliconANGLE/Microsoft Designer