Ollama offers Reflection's Beam free for a limited time
Reflection has released its first open-weight model, and it is aiming straight at developers who build agents that work for hours, not seconds. Available through Ollama Cloud, Beam arrives with a privacy promise that intrigues anyone tired of trading security for power: what happens to your code once the model has seen it?
Ollama offers Reflection's Beam free for a limited time
Ollama is a day-zero partner, rolling out Beam for free for a limited time. In collaboration with Reflection, Ollama's Beam endpoint runs with full zero data retention (ZDR). You can reach Beam through Ollama Cloud, where you can run supported models from your apps or terminal without downloading models or managing local inference.
Beam is a mixture-of-experts model with approximately 501 billion total parameters and 23 billion active parameters per token. It is designed for software engineering, reasoning, and long-horizon agentic tasks. Its text-in, text-out design targets workflows that involve more than a single response, such as writing and maintaining code, using tools across multiple steps, and working through complex tasks over time. In these workloads, code, tool calls, logs, documents, and structured outputs are central. Beam has a 1 million token context window to support long-running workflows.
Beam's uses include coding agents that solve software engineering tasks and work through codebases, long-running agents that execute multi-step workflows and use tools toward a goal, math and science reasoning, and custom models that you can fine-tune for specific applications. The article includes benchmark charts for agentic coding, reasoning, tool calling, and general performance, but it does not state scores in the text.
Key Features at a Glance
Open weights: Released under Apache 2.0, so developers can inspect, customize, and build on the model.
Scale: Approximately 501 billion total parameters, with 23 billion active per token, using a mixture-of-experts design.
Context window: 1 million tokens for long-running workflows.
Agentic focus: Built for multi-step coding, tool use, and extended task sequences.
Privacy: The Ollama Beam endpoint runs with full zero data retention (ZDR).
No local setup: Run through Ollama Cloud without downloading models or managing local inference.
Availability
Beam is generally available as of October 16, 2026, through Ollama Cloud, free for a limited time. To get started, download the latest version of Ollama, then run one of these commands:
