Reflection's open 501B Beam model is free on Vercel AI Gateway with 1M context
Reflection open-sourced Beam 1.0, a 501B-parameter model, on October 16, and Vercel AI Gateway added a free version as reflection/beam-free with $0 input and output. It has about 1M tokens of context and up to about 131K output tokens, is built for coding and agentic tasks, and has no published usage cap or end date.
The briefing
Reflection open-sourced its first model, Beam 1.0, on October 16, 2026, and Vercel AI Gateway added a free version, Beam 1.0 (Free). The model ID is reflection/beam-free, and both input and output are billed at $0.
What you can use
Beam is a mixture-of-experts model with 501B total parameters and 23B active per token, built for coding, reasoning, and agentic work. The weights are on Hugging Face under the Apache 2.0 license.
The free version has a context window of about 1M tokens and up to about 131K output tokens, with reasoning tokens counted as output. It supports tool calling and structured outputs that follow a JSON schema, and reasoning effort can be set to low, medium, high, xhigh, or max. It accepts text input only.
Beam is free for a limited time during its launch period, and Vercel has not published a usage cap or end date. There are two free routes. The standard model name reflection/beam with provider reflection is free now and switches to regular pricing automatically when the free period ends, with no config change or downtime. The reflection/beam-free slug stops serving when the offer ends, which suits you if you never want to be charged.
The free Reflection provider does not offer Zero Data Retention, and prompts and responses may be used for training. reflection/beam also has paid providers: Nebius at $0.25 per million input tokens and $1 per million output tokens, Together AI at $0.40 and $1.20, and Baseten at $1.30 and $4.30.
Reflection's own API is also in a free public beta with a daily per-account allowance that resets at midnight Pacific Time, but new sign-ups join a waitlist before they can create API keys. On OpenRouter, Beam is paid only for now, at $0.25 per million input tokens and $1 per million output tokens.
How to use it
- Sign in to Vercel and create an AI Gateway API key for your team.
- Call it through the AI SDK, the OpenAI-compatible Chat Completions and Responses APIs, or the Anthropic-compatible Messages API, with model
reflection/beam(providerreflection) orreflection/beam-free. - To use it in Claude Code, Codex, or Cursor, install the latest Vercel CLI, run
npx vercel ai-gateway setup, then pick reflection/beam or reflection/beam-free in your agent.
More details: Vercel changelog · Beam 1.0 (Free) model page · Hugging Face launch post.