Reflection Beam 1.0
Reflection's open-weight reasoning model for coding and long-horizon agentic work: a sparse mixture of experts with 501B total and 23B active parameters, a 1M-token context and tool calling.
Reflection's Beam 1.0 is a cloud text model. It takes a text prompt and replies with generated text. Its context window handles up to 1.05M input tokens and up to 131K output tokens, and it supports adjustable reasoning effort for harder problems. Generation can be tuned with system instructions, temperature, and top-p sampling. It runs through OpenRouter using your own API key, from $1.00 per million input tokens.
- Released
- Oct 2026
- Pricing
- $1.00 / 1M in · $3.00 / 1M out
- Context window
- 1.05M in · 131K out
- Reasoning
- Adjustable effort
- Controls
- System prompt, Temperature, Top-p
Samples
Generated with Beam 1.0 via OpenRouter. The same three prompts run against every text model in the catalog, shown verbatim, so the only thing that changes between two models’ answers is the model.
Explain what an API is to someone who has never written code. Do not use a restaurant or waiter analogy. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 50 words.
An API is a set of rules that lets one piece of software ask another program to do work or share data. It acts like a connector, so developers can use features without knowing the full internal code.
A two-person startup is deciding whether to build their own user authentication or use a hosted provider. Give them a clear recommendation, then the single strongest argument against your own recommendation. Plain prose only — no headings, bullet points, or markdown formatting. Maximum 150 words.
Use a hosted authentication provider. It lets a two-person team avoid building and maintaining password storage, multi-factor flows, account recovery, and breach monitoring, while benefiting from security expertise and faster shipping. The strongest argument against this is that outsourcing a critical security boundary creates vendor lock-in and reduces direct control over user data, privacy, compliance, and custom authentication behaviors.
Write a briefing for a non-technical executive explaining why their company's spending on AI is hard to predict, and what they can do about it. Cover at least three distinct causes. Plain prose only — no headings, bullet points, or markdown formatting. Write exactly 205 words.
Your AI spending is hard to predict because token-based pricing grows with every added word, conversation turn, and tool call, so a successful feature can quietly multiply usage. It is also hard because vendors regularly change model prices, rate limits, and usage terms, while internal teams experiment with stronger models before knowing which simple ones would suffice. A third reason is that direct API fees are only part of the bill: evaluation, data labeling, security review, guardrails, observability, and integration maintenance can become large line items as deployments scale. You can reduce uncertainty by approving an AI budget with weekly alerts, routing routine tasks to lower-cost models, caching repeated answers, capping request frequency, and requiring owners to forecast expected traffic before launch. You should also renegotiate vendor commitments, compare providers quarterly, and track value metrics such as cases resolved or revenue lifted rather than model output alone. Teams should pilot with logs, define stop-loss rules, and retire unused agents because abandoned experiments often continue consuming tokens. Finance and product leaders should review billing reports jointly each month, separating experimentation from production so one budget does not hide the other. With this discipline, spending becomes a managed operating expense instead of an unwelcome surprise each quarter.
Reflection
Reflection AI builds open-weight frontier models, including Beam, a reasoning model for coding and long-running agents.
reflection.ai ↗