Skip to content

Beam 1.0

Reflection AI

Key facts · verified Oct 27, 2026

Beam 1.0 is a Reflection AI model released on 5 Oct 2026. Pricing: Not published — open weights (Apache 2.0); Reflection API. Its context window is 1M tokens. It is not yet ranked on LMArena's text leaderboard. Source: huggingface.co.

Reflection AI’s Beam 1.0, its first open-weight model, announced on 5 October 2026: a sparse mixture-of-experts model with 501 billion total and 23 billion active parameters and a 1-million-token context window, built for coding, reasoning and agentic tasks. It takes and returns text, ships in BF16, FP8 and NVFP4, and is released under Apache 2.0. Reflection serves it through its own API and its Mirror coding agent; it has not published a per-token price.

Pricing & specs checkedSource: vendor documentation · checked Oct 27, 2026
Pricing

Not published — open weights (Apache 2.0); Reflection API

Context

1M tokens

LMArena Elo

Not rated

LMArena rank

Not ranked

Key Features

Open Weights Coding Agentic Tool Use Long Context Self-Hostable

Full Specifications

Not published — open weights (Apache 2.0); Reflection API
Identity
DeveloperReflection AI
ReleasedOct 2026
LicenceApache 2.0 (open weights)
Self-hostableYes — BF16, FP8 and NVFP4 weights
Capacity
Context window1M tokens
Capability
Vision inNo — text in, text out
Access
APIYes — Reflection API (OpenAI SDK compatible)
Fine-tuningYes — Reflection documents fine-tuning and post-training with open-source stacks