Announced October 5, 2026, Beam is Reflection's first open-weight release — a 501-billion-parameter sparse mixture-of-experts model for coding, reasoning, and agentic workloads, with weights set to ship under an Apache 2.0 license later this month.
2026-10-05 · by Moe Ameen
Reflection AI introduced Beam on October 5, 2026 — the company's first open-weight model. Beam is a text-only sparse mixture-of-experts (MoE) system with 501 billion total parameters and about 23 billion active per token, a context window Reflection lists at 1 million tokens, and a clear focus: coding, reasoning, and agentic workloads such as software engineering, terminal operations, tool use, and web search. Reflection frames the release around "frontier open intelligence" and positions Beam to compete with the leading open-weight models, including the Chinese releases that have set the pace this year. Reflection has been reported as a Nvidia-backed startup.
On the training side, Reflection says Beam was pretrained on 23.8 trillion tokens and then put through a large reinforcement-learning run — more than 100 million rollouts across roughly 10,500 NVIDIA GB300 GPUs over about four weeks. The company reports strong figures on coding and reasoning benchmarks (including SWE-Bench Pro, Terminal-Bench, AIME 2026, and GPQA Diamond) and claims Beam matches Z.ai's GLM-5.2 on advanced reasoning while using three to four times less inference compute. Those numbers are Reflection's own and had not been independently verified at announcement.
The weights were not available on day one. Beam was announced with early access via a waitlist while the model finishes final red-teaming and evaluations; Reflection said it will release the weights under an Apache 2.0 license later in October 2026, alongside a technical report, a model card, and the full stack for running, evaluating, and fine-tuning the model. Confirm the exact spec, benchmark numbers, and release terms on Reflection's own pages before relying on them, since early launch details move.
The real signal here is not one model's benchmark — it is that frontier-scale reasoning keeps shipping as open weights you can run and fine-tune yourself. That is genuinely good for creators and creator-developers, and it changes where the bottleneck sits. A model like Beam can plan a content calendar, run agentic research over a topic, or draft a script from a messy source — but it stops at correct text on a screen. It puts no face on the idea, builds no scroll-stopping format, and gets it in front of no one. The downstream half is where a content operation actually lives, and it is the half a reasoning model does not touch.
That is the job [Kompozy](/) is built for, and it is deliberately model-agnostic about the brain upstream — its own copy runs on Claude and OpenAI, governed by a written [Persona Brief](/glossary/persona-brief), so you never wrangle a model to get on-brand output. Take the plan or draft (from Beam, from any model, or from a raw source) and the engine generates across 18 formats: a [Persona Short](/glossary/persona-shorts) that narrates the idea on camera with a face-locked avatar, a brand-exact [Carousel](/glossary/hyperframes) that breaks down a complex point, Quote Graphics, plus a Blog Article and an Email Newsletter — then [Autopilot](/glossary/autopilot) reframes each asset per platform and publishes across the eight social platforms plus blog and email behind a per-post review. Open-weight models make the thinking cheap; Kompozy turns the thinking into finished, on-brand content that actually ships.
Beam is Reflection AI's first open-weight model, announced October 5, 2026. It is a text-only sparse mixture-of-experts model with 501 billion total parameters (about 23 billion active) and a 1M-token context, built for coding, reasoning, and agentic workloads. Reflection says it will release the weights under an Apache 2.0 license later in October 2026.
Reflection describes Beam as open-weight and says it will publish the weights under an Apache 2.0 license later in October 2026, along with a technical report, a model card, and tools for running, evaluating, and fine-tuning it. At announcement, access was via an early-access waitlist while the model finished red-teaming. Confirm the exact license and timing on Reflection's site.
No. Beam is a text-only model — it writes text and code and reasons over inputs, but it generates no images, video, audio, captions, or published posts. To turn its output into finished, multi-format content you pair it with a generation-and-publishing engine like Kompozy.
Treat it as another cheap, capable reasoning brain rather than a content tool. Use Beam (or any model) to research, plan, and draft, then hand the text to a content engine. Kompozy takes a draft or source and produces persona/avatar video, carousels, images, a blog, and a newsletter in one brand voice, then schedules and publishes across eight social platforms plus blog and email.