Ollama:欢迎登上开源模型列车
2026 年 7 月 9 日
迈克尔和我在大学相识,在那里我们创办了第一家公司 Kitematic,它让 Docker 的运行变得极其简单。2015 年,该公司被 Docker 收购。在那里,我们的工作成果变成了 Docker Desktop,于 2016 年推出,如今已被全球超过一千万开发者使用。
十年后,我们回来了,正在打造 Ollama——让开发者能够最轻松地启动并运行开源模型。出乎我们最疯狂的想象,Ollama 已迅速成为访问开源模型的主要平台,服务着 890 万开发者,他们和我们一样,相信 AI 应该由你来构建、运行和打造。
AI 的个人电脑时刻
计算领域曾经经历过这样的时刻:个人电脑将机器从大型机机房中取出,放到你的桌面上——供你拥有、定制和构建。开源模型如今正在为 AI 开启这样的时刻。
几年前,我们目睹了强大的新型开源模型的出现。它们可以免费下载,但要让它们正常工作却很困难。能力是存在的,但对于那些希望像通过 API 访问专有模型一样使用它们的开发者来说,这些能力尚未被解锁。
于是我们开始行动。我们的团队推出了 Ollama,这是一款可以下载到电脑上的应用,只需一条命令就能启动并运行最新的开源模型,并通过简单的 API 在其基础上进行构建。Ollama 迅速成为开发者们能够自如地发现最新开源模型并学习如何将其投入使用的家园。
有了 Ollama,运行开源模型变得像运行任何其他软件一样简单:无需权限、API 密钥或昂贵的服务器硬件。你的模型。你的机器。你的数据。
你拥有的 AI:个人化且私密
我们围绕对客户和社区最重要的三个原则构建了 Ollama:
所有权。开源模型归你所有,你可以保留、定制和优化。你永远不会被锁定在你智能体或应用所依赖的模型之外,并且拥有完全的自由来更改它们。
可负担性。在你自己的硬件上运行的模型不会带来失控的按 token 计费账单。你可以进行实验、迭代和发布,而无需担心每次提示词都会增加成本。
隐私。使用可在本地运行的开源模型,你的数据永远无需离开你的机器。当你确实需要扩展规模时,也能将同样的信任带入云端。
扩展开源模型,解决世界上最棘手的问题
开源模型不再是实验或研究项目。它们如今正运行在全球最大的公司内部——Ollama 已被 85% 的《财富》500 强企业使用——而这仅仅是个开始。
最初只是在个人电脑上首次运行开源模型的乐趣,如今已演变为解决那些曾专属于闭源模型的极困难问题的动力。Ollama 云是访问最强大开源模型的最便捷方式:GLM、Nemotron、DeepSeek、Kimi、MiniMax 等。平均而言,Ollama 云的 token 量每月增长超过一倍,我们迫不及待地想让每个团队都能使用它。
全力押注开源
今天,我们宣布 Ollama 已从 Benchmark 的 Peter Fenton、Theory Ventures 的 Tomasz Tunguz、8VC 的 Alex Kolicich,以及 Solomon Hykes(Docker 创始人)、Aaron Katz(ClickHouse CEO)、Spencer Kimball(GIMP 联合创建者及 Cockroach Labs 联合创始人)、Quinn Slack(Amp CEO)、Marianna Tessel(思科董事会成员)、Michael Montano(Twitter 前工程主管),还有 Y Combinator、Garage Capital、Pace Capital、49 Palms、GTMFund 以及众多其他出色的天使投资人处筹集了 8800 万美元。
这笔资金是未来发展的燃料。Ollama 处于开源模型生态系统的核心位置,我们打算尽全力推动其发展:无缝的混合推理、在新开源模型发布当天即提供支持,以及一个能让任何开发者及其团队使用最强大模型而无需放弃所有权或隐私的云服务。
这一切都离不开模型背后的团队。感谢那些发布强大、可公开获取的模型并推动整个领域前进的团队。我们的工作是让每个人都能轻松运行他们出色的成果,并保持其开放性。
我们曾站在类似变革的开端,押注开放和易用会胜出。如今我们再次倾尽所有,押注于此。
是时候推动开源人工智能向前发展了。全体上车。
要开始使用,请下载 Ollama 并开始用开源模型进行构建:
ollama
感谢每一位运行过模型、提交过问题、构建过集成或用 Ollama 发布过作品的人。我们非常乐意听取你们的反馈。我们能走到今天,全靠你们。
——杰夫与迈克尔
Ollama: all aboard open models
July 9, 2026
Michael and I first met in college, where we started our first company, Kitematic, which made Docker dead-simple to run. In 2015, it was acquired by Docker. There, our work became Docker Desktop, launched in 2016 and now used by over ten million developers around the world.
Ten years later, we’re back, building Ollama — the easiest way for developers to get up and running with open models. Beyond our wildest dreams, Ollama has quickly become the leading platform to access open models, serving 8.9 million developers who, just like us, believe that AI should be yours to build, run, and make your own.
The personal computer moment for AI
Computing has been here before: the personal computer took the machine out of the mainframe room and put it on your desk — yours to own, customize, and build with. Open models are now enabling that moment for AI.
A few years ago we watched as powerful new open models emerged. They were freely available to download, but it was hard to get them working. The power was there, but it wasn’t unlocked for developers who wanted to consume them the way they could access proprietary models behind an API.
So we got to work. With our team we shipped Ollama, an app you could download on your computer to get up and running with the latest open models in a single command, and build on them via a simple API. Ollama quickly became the place where developers could feel at home to discover the latest open models and learn how to put them to work.
With Ollama, running an open model became as easy as running any other piece of software: no permission, API key, or expensive server hardware required. Your model. Your machine. Your data.
AI that you own: personal & private
We built Ollama around three principles that matter most to our customers and community:
Ownership. Open models are yours to keep, customize, and optimize. You’re never locked out of the models your agent or app depends on, and you have complete freedom to change them.
Affordability. Models that run on your own hardware don’t come with runaway per-token bills. You can experiment, iterate, and ship without worrying that every prompt is adding cost.
Privacy. With open models that can run locally, your data never has to leave your machine. When you do need to scale, you can carry that same trust into the cloud.
Scaling open models to solve the world’s hardest problems
Open models aren’t an experiment or a research project anymore. They’re now running inside the largest companies in the world — Ollama is used by 85% of the Fortune 500 — and it’s just the beginning.
What started as the joy of running an open model on your computer for the first time has evolved into the drive to solve very difficult problems once reserved for proprietary models. Ollama’s cloud is the most delightful way to access the most powerful open models: GLM, Nemotron, DeepSeek, Kimi, MiniMax and more. On average, Ollama’s cloud has more than doubled in token volume every month, and we can’t wait to make it available to every team.
All in on open
Today, we’re announcing that Ollama has raised $88M from Peter Fenton at Benchmark, Tomasz Tunguz at Theory Ventures, Alex Kolicich at 8VC, along with Solomon Hykes (founder of Docker), Aaron Katz (CEO of ClickHouse), Spencer Kimball (co-creator of GIMP and co-founder of Cockroach Labs), Quinn Slack (CEO of Amp), Marianna Tessel (board of directors, Cisco), Michael Montano (former head of engineering at Twitter), and Y Combinator, Garage Capital, Pace Capital, 49 Palms, GTMFund, and many other incredible angel investors.
This funding is fuel for what’s ahead. Ollama sits front and center in the open model ecosystem, and we intend to push it forward as hard as we can: seamless hybrid inference, support for new open models the day they’re released, and a cloud that lets any developer and their team reach the most powerful models without giving up ownership or privacy.
None of this happens without the people behind the models. Thank you to the teams releasing powerful, openly available models and moving the entire field ahead. Our job is to make their incredible work effortless to run for everyone and to keep it open.
We’ve stood at the beginning of a shift like this once before, betting that open and easy would win. We’re betting on it again, with everything we’ve got.
It’s time to drive open-source AI forward. All aboard.
To get started, download Ollama and start building with open models:
ollama
Thank you to everyone who’s run a model, filed an issue, built an integration or shipped something with Ollama. We love hearing your feedback. We’re only here because of you.
— Jeff & Michael