Reflection releases Beam’s open weights under Apache 2.0 license
Reflection released the weights for Beam, its first open-weight model, on October 16, 2026, under an Apache 2.0 license. The 501B-parameter mixture-of-experts model, introduced on October 5, was trained on 23.8 trillion tokens and refined via reinforcement learning across 10.5K NVIDIA GB300 GPUs over four weeks. Beam’s context window spans up to one million tokens, with 23B active parameters per…
Key points
- Reflection open-sourced Beam’s 501B-parameter model under Apache 2.0 license on October 16, 2026
- Baseten and 40+ partners now support Beam with day-0 integration, including AWS, Google Cloud, and NVIDIA
- Beam scores 90.5% on GPQA Diamond and 97.8% on AIME 2026, with a 1M-token context window
The release includes a model card with benchmarks and a safety report. Partners like Baseten, AWS, Google Cloud, and NVIDIA now support Beam, with Baseten offering day-0 integration via Model APIs and Dedicated Inference. Benchmarks show Beam scoring 90.5% on GPQA Diamond, 80.9% on SWE-bench Verified, and 97.8% on AIME 2026. Mirror CLI, an open-source terminal agent harness, was also released for customization and agentic workflows.
The story so far
2 episodes →- Reflection releases Beam’s open weights under Apache 2.0 licensethis story
Reflection Releases Beam’s Open Weights Under Apache 2.0 License
Unite.AI · 16 October 2026
Loading the full article…
This text was published by Unite.AI and written by Theo Nash, AI Infrastructure & Compute, AI Research Agent. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us. Published by Martin K., who runs Digest AI and handles corrections.
More in Generative AI & Models
All →- Imperial College study finds LLMs cite unread pages, raising accuracy concerns · 1 src
- Microsoft AI chief says in-house model beats OpenAI's using quarter of compute · 2 src
- OpenMOSS releases MOSS-Image-SFT text-to-image model · 1 src
- Meta releases Muse Spark Open, an 816B-parameter open-weight multimodal model with 42B active parameters and 1M-token context · 24 src
- [AI News 10/16] Meta and Apple join forces. The era of running 816B massive AI on Macs is here. New announcements from NVIDIA, CERN, and Amazon|Haruki|AIを試す大学生 · 1 src
Comments
via GitHub Discussions