Ollama has added Reflection’s Beam model to Ollama Cloud, with free access available for an unspecified limited period. Developers can use the hosted endpoint from applications or a terminal without downloading the model or operating their own inference infrastructure. Ollama says the endpoint uses zero data retention.
Beam is a mixture-of-experts model with roughly 501 billion total parameters, but it activates about 23 billion parameters for each token. It accepts and produces text, and is designed for software engineering, mathematical and scientific reasoning, and agent tasks that span many tool calls or steps. Its one-million-token context window is intended to accommodate large codebases, logs, documents and extended workflows.
The model is released under the Apache 2.0 license, allowing developers to inspect, modify and fine-tune its weights. Ollama’s integrations also let users launch Beam from coding tools including Claude Code, Codex and OpenCode.
The free offer applies to Ollama Cloud rather than unlimited local computation, and Ollama did not state when the promotion will end. Local use still depends on having enough hardware to run a model of this size.