Benchmark leads a major funding round for the open-source tool that lets developers run AI models directly on their own hardware
Running a large language model used to require either a corporate budget or a PhD in cloud infrastructure. Ollama changed that. The open-source tool lets developers spin up AI models on their own laptops with roughly the same effort as installing a podcast app.
Ollama has raised $65 million in a funding round led by Benchmark, as the company looks to scale its developer platform to 9 million users.
What Ollama actually does #
The tool allows developers to download and run open-source models, such as Meta’s Llama series or Mistral’s releases, directly on their own machines. No API keys, no usage bills, no sending your data to a third party.
The company’s GitHub repository has accumulated approximately 176,000 stars and around 16,900 forks. Monthly downloads reached 52 million as of early 2026.
Why Benchmark is paying attention #
The competitive landscape here is real. Tools like vLLM and Apple’s MLX framework serve overlapping audiences, and each has carved out meaningful adoption among developers building inference pipelines. Ollama’s edge has been simplicity: a clean command-line interface and a library of pre-packaged models that work out of the box on consumer hardware.
The crypto angle: local AI meets onchain agents #
The platform has been integrated into onchain AI agent projects, with Shinkai being a notable example. Shinkai uses Ollama to power AI agents that can execute tasks and process USDC payments on-chain.
Ollama does not have its own token and has no direct blockchain integration at the protocol level. But its role as infrastructure for projects that do operate on-chain gives it indirect relevance to the broader decentralized AI narrative.
What this means for developers and investors #
For investors tracking the AI infrastructure space, Ollama’s round is a data point in a broader argument that the value in AI tooling does not accrue exclusively to companies running massive GPU clusters. Local inference has real advantages in cost, latency, and data privacy, and those advantages become more pronounced as capable open-source models become easier to run on consumer hardware. Disclosure: This article was edited by Editorial Team. For more information on how we create and review content, see our