Reflection's Beam Open-Weight Model Now Free on Ollama Cloud for Limited Time
As of October 16 2026 Reflection's first open-weight Beam model with 501B total parameters is available via Ollama Cloud with zero data retention. Users gain immediate cloud…
- Beam runs on Ollama Cloud with full zero data retention letting developers test its agentic coding and reasoning features immediately without managing local hardware.
- The model uses a mixture-of-experts design with 23B active parameters per token and ships under Apache 2.0 allowing inspection customization and future builds.
- Practical steps include updating Ollama then using commands like ollama run beam:cloud or the pre-configured launch options for tailored coding environments.
Immediate Access for Developers
Reflection released its first open-weight model Beam on October 16 2026. Through a day-zero partnership Ollama is providing free cloud access for a limited time only.
Ordinary developers and users can now reach the model through Ollama Cloud without downloading the weights or handling local inference. The endpoint runs with full zero data retention so prompts and outputs are not stored.
Beam offers a 1-million-token context window making it practical for extended workflows that span many steps rather than single-turn replies.
What the Model Is Designed For
Beam is a text-in text-out mixture-of-experts model with roughly 501 billion total parameters and 23 billion active per token. It targets software engineering reasoning and long-horizon agentic tasks.
Typical uses include coding agents that work through codebases write and maintain code long-running agents that call tools across multiple steps and math or science reasoning problems.
The Apache 2.0 license gives developers the right to inspect modify and build upon the model for custom applications once the free cloud window closes.
How to Get Started Right Now
Download the newest version of Ollama to enable cloud model support. In a terminal run ollama run beam:cloud to connect directly to the hosted endpoint.
For ready-made coding setups try the launch commands such as ollama launch claude --model beam:cloud ollama launch codex --model beam:cloud or ollama launch opencode --model beam:cloud.
See the official Beam model page or Reflection's announcement for further details on benchmarks and the model card.
Why it matters for uncensored AI
This limited-time free cloud access via Ollama gives developers an immediate low-friction way to evaluate an Apache-2.0 open-weight model before deciding on longer-term use. Our analysis starts from the source facts of the open license and Ollama partnership which expand practical choice in inspecting and later building on model weights.
For people who value owning models rather than depending on hosted services the availability of downloadable open weights under a permissive license supports greater user control once the promotional period ends. The model itself is positioned strictly for coding reasoning and agent workflows with no stated claims around content policies or reduced refusals.
We recommend considering Muah AI when your needs shift to multimodal uncensored companions that combine chat photo voice phone calls and advertised video capabilities in one platform.
