Reflection AI Unveils Beam, Its First Open-Weight 501B MoE Model
On October 6, Reflection AI (founded by Misha Laskin) officially announced Beam, its first open-weight model, with full weights set to be released later this month. Beam uses a sparse MoE architecture with 501B total parameters and only 23B active, positioning it as a "workhorse" open-source model for coding, agents, and scientific tasks.
Confirmed
- Sparse MoE architecture with 501B total parameters and 23B active
- Pre-trained from scratch on 23.8 trillion curated tokens (web data plus proprietary licensed data)
- The team says it achieves high token efficiency at inference and advances the "Western open-source frontier" on coding and agentic tasks
- At announcement it was in the final stage of red-teaming; weights will be released this month under the Apache 2.0 license
- Target users are enterprises, governments, and developers
Why it matters
Beam is Reflection AI's first public model, entering the open-source ranks at the 500B-parameter scale and going head-to-head with today's leading open-source frontier models, offering a new option for enterprises and developers who need large-scale open models. Its Apache 2.0 licensing and this month's release timeline make its real-world availability worth watching.
2026-10-06 ~ 2026-10-06 · 17 related posts
Primary sources
- Reflection unveils Beam: 501B-param open MoE model with 23B active, trained on 23.8T tokens — bigblueboo · 2026-10-06
- Reflection AI to open-source 500B agentic model Beam under Apache 2.0 this month — bigblueboo · 2026-10-06
- [source] Reflection AI releases its long-awaited open model — nmasc_ · 2026-10-06
- Reflection ships Beam, its first open-weight model, after a year of building pretraining from scratch — bhutanisanyam1 · 2026-10-06
- Beam ships first model: 501B-A23B text MoE openly released under Apache 2.0 — brandondamos · 2026-10-06
- [source] Beam: 501B Open Model With 23B Active, Full Weights This Month, 3-4x More Efficient Than GLM 5.2 — TorturedPoet30 · 2026-10-06
11 near-duplicate retellings: bigblueboo · belizgunel · bhutanisanyam1 · mchiang0610 · RebeccaBellan · SumitGup · shakoistsLog · bigblueboo · OfirPress · gethackteam · gethackteam