RT Amir Haghighat：“Baseten何时提供沙盒？”一直是客户常见的问题。而我们已迈出了巨大的一步来回答它：Baseten ha...
RT Amir Haghighat: “When will Baseten offer sandboxes” has been a common customer question. And we’ve taken a huge step in answering it: Baseten ha...
Aditya AgarwalAI2026-09-10
今天，Blaxel加入Baseten。通过强强联合，我们旨在为下一个万亿级智能体构建云平台。Baseten已构建了团队用于训练和部署模型的领先性能、可靠性和效率的基础设施。Blaxel则构建了智能体需要的状态执行层：快速、隔离的沙箱、持久存储和生产级连接。携手并进，我们将Baseten扩展为基础设施基础，团队可以在其中部署和训练模型，现在还可以在一个集成平台上构建长期运行的智能体。两家公司都坚信，人工智能将成为无处不在的技术，并需要新一代基础设施。Baseten的创立基于这样一个信念：在规模上提供快速、可靠的推理将成为人工智能应用的瓶颈。Blaxel的创立则基于这样一个信念：现有的云平台并未为人工智能智能体的运行方式设计，也无法在规模上支持它们。在过去的18个月里，我们在Blaxel上构建了这一新的云平台，一次构建一个基础组件。沙箱是最先出现的，因为没有它们，其他任何东西都无法工作。智能体编写和运行代码，调用工具，并在长时间任务中携带状态，而在共享进程中这样做是不安全的。每个智能体都拥有自己的微虚拟机。我们使它们非常快速：挂起和恢复只需25毫秒，比其他沙箱产品快5倍。沙箱可以闲置数月，几乎零成本，并在模型完成下一句话之前恢复，这样状态就可以持续，而无需为未使用的计算付费。Sapiom在Blaxel上运行了数亿个智能体循环。正如那里的创始人工程师David Zhang所说：“Blaxel以一种我通常预期需要一年或更长时间才能达到的方式处理了这种增长。”智能体还需要持久化和共享许多工件：文件、代码和工作上下文。这些必须比单次运行更持久，并且可以供下一个沙箱或智能体访问。智能体驱动是我们的分布式...
原文
Today, Blaxel is joining Baseten. By joining forces, we aim to build the cloud for the next trillion agents.Baseten has built the infrastructure that teams use to train and serve models with industry-leading performance, reliability, and efficiency. Blaxel has built the stateful execution layer agents need to act: fast, isolated sandboxes, persistent storage, and production-grade connectivity. Together, we are expanding Baseten into the infrastructure foundation where teams can serve and train models and can now build long-running agents, all in one integrated platform. Both companies were founded with the conviction that AI would become ubiquitous and require a new generation of infrastructure. Baseten was started with the belief that delivering fast, reliable inference at scale would become a defining bottleneck to AI adoption. Blaxel was founded on the belief that existing clouds weren't designed for how AI agents run and couldn't support them at scale.For more than 18 months, we at Blaxel have been building this new cloud, one primitive at a time.Sandboxes came first because nothing else works without them. Agents write and run code, call tools, and carry state across long tasks, and you can't do that safely in a shared process. Every agent gets its own microVM. We made them extremely fast: suspend and resume in 25 milliseconds, up to 5x faster than other sandbox products. A sandbox can sit idle for months at close to zero cost and come back before the model finishes its next sentence, so state persists without paying for compute you aren't using.Sapiom runs hundreds of millions of agent loops on Blaxel. As David Zhang, founding engineer there, put it: "Blaxel handled that kind of growth in a way I'd normally expect to take a year or more to reach."Agents also need to persist and share many artifacts: files, code, and working context. That has to outlive a single run and be accessible to the next sandbox or agent. Agent Drive is our distributed filesystem for that. Mount it into any sandbox and whatever an agent writes is durable, versioned, and there when something else needs it. And agents have to talk to things. Tools, MCP servers, APIs, other agents. We rebuilt a networking layer that connects all of these pieces with the isolation and access controls a production team will actually sign off on. The primitive we didn't own was inference. We've seen open-weight & custom models become core to how autonomous agents get deployed in production. Inference is no longer a service you call from another data center. It has to sit next to compute, storage, and network; or latency compounds on every loop. That brings us to Baseten – joining forces is how we colocate all four, bringing the brain next to the muscles. Baseten's inference stack is the best on the market, running at scale across dozens of regions, with more than $2B raised to build it. Our shared vision is a future where many agents use many models