Editorial illustration for Reflection's 501B Beam model is free on Ollama for a limited time
Reflection's Beam Model Free on Ollama Now
Reflection released Beam, its first open-weight model, on October 16, 2026, and Ollama is rolling it out as a day-zero partner, free to use for a limited time. The model runs on Ollama Cloud, meaning developers can call it from apps or a terminal without downloading weights or managing local inference themselves. Reflection and Ollama also built the Beam endpoint with full zero data retention, so nothing gets stored on the way through.
Under the hood, Beam is a mixture-of-experts model with roughly 501 billion total parameters, though only about 23 billion are active per token. That architecture is aimed squarely at software engineering, multi-step reasoning, and agentic tasks that stretch across many actions rather than a single exchange. Reflection licensed it under Apache 2.0, so developers can inspect the weights, modify them, or fine-tune the model for their own purposes.
The pairing of a large reasoning-focused model with a free, no-download cloud endpoint is the headline here. What Beam is actually built to do, and the kinds of workflows Reflection designed it around, is where the details get specific.
In collaboration with Reflection, Ollama’s Beam endpoint runs with full zero data retention (ZDR). You can access Beam through Ollama Cloud, where you can run supported models from your apps or terminal without downloading models or managing local inference.
Why this matters
A 501-billion-parameter model with 23 billion active per token, running free on Ollama Cloud with zero data retention, is a real chance to test a frontier-scale agentic model without the usual API meter running or the privacy trade-offs that come with it. For developers and founders, that's worth a weekend, not a shrug. Beam is pitched at software engineering and long-horizon agentic tasks, meaning multi-step workflows like maintaining a codebase over time rather than one-shot Q&A, which is exactly the use case most teams are still trying to figure out how to trust a model with.
The "limited time" framing matters too: this is a promotional window, not a permanent pricing tier, so anyone building against it should assume costs show up eventually. We'd treat this as a window to stress-test Beam's actual reasoning and agentic chops against Reflection's claims before the free access disappears, not as evidence that 500-billion-parameter open-weight models are now just free infrastructure.
Related Reading
LLMs & Generative AI ChatGPT's 'Nerdy' tweak rewards goblin metaphors in answers, study finds
LLMs & Generative AI Google to End Free Access to Gemini AI for Non-Subscribers
LLMs & Generative AI AI Engineers Face Rising Costs, Need New Strategies for Efficiency
From Archives Liquid AI releases multimodal d1-ultra, scores 65 on AA Decision Index
From Archives Feature Elicitation for SAE Steering Reveals Deceptive Defanging in Small LLMs
Common Questions Answered
What is Reflection's Beam model and how does it differ from other open-weight models?
Beam is Reflection's first open-weight model released on October 16, 2026, featuring a 501-billion-parameter architecture with 23 billion active parameters per token. It is specifically designed for software engineering and long-horizon agentic tasks involving multi-step workflows, distinguishing it from general-purpose models by its focus on complex, sustained problem-solving scenarios.
How can developers access Beam through Ollama without managing local inference?
Developers can access Beam through Ollama Cloud, which allows them to call the model directly from apps or a terminal without downloading model weights or managing local inference infrastructure. This day-zero partnership between Reflection and Ollama provides free access for a limited time, making frontier-scale models immediately available to developers.
What does zero data retention mean for the Beam endpoint on Ollama Cloud?
Zero data retention (ZDR) means that nothing gets stored as data passes through the Beam endpoint, ensuring complete privacy and eliminating the typical privacy trade-offs associated with using cloud-based APIs. This feature allows developers to test a frontier-scale agentic model without concerns about their data being retained or monitored.
Why is free access to Beam on Ollama significant for developers and founders?
A 501-billion-parameter model running free on Ollama Cloud with zero data retention provides a rare opportunity to test frontier-scale agentic capabilities without API costs accumulating or privacy concerns. This removes traditional barriers to experimentation with advanced models, making it worthwhile for developers to explore Beam's potential in multi-step workflows and complex software engineering tasks.
