Beam: Reflection AI’s open 501B model under Apache 2.0
Reflection AI has published the weights of Beam, a 501B-parameter model, on Hugging Face under Apache 2.0, which also allows commercial use.
Source of truthhuggingface.co/blog/reflection-beamOn October 16, 2026, Reflection AI released Beam’s weights on Hugging Face under the Apache 2.0 license. The model card says it is intended for commercial and research use and does not prohibit fine-tuning or quantization. It has 501B parameters (23B active), handles text only, has a 1M-token context window and was developed mainly for English. In NVFP4 the weights take about 280 GB. According to Reflection, it is built for coding and long-horizon agentic work and scores 81.6% on SWE-bench Verified in BF16. According to Reflection’s announcement, its own API is in public beta and free to start; the announcement gives no prices.
Other sources: huggingface.co/…/Beam-1.0-501B-A23Breflection.ai/blog/beam-is-available
Related stories
Granite Speech 5.0: IBM releases Apache 2.0 speech models
IBM released granite-speech-5.0-2b and revised two more speech models, all under the Apache 2.0 license. Greek is not listed for the 2B and 470M models.
Black Forest Labs: FLUX 3 Image Klein, an open-weights image model
An open-weights model for creating and editing images; the INT8 build runs on GeForce RTX cards with 8GB of VRAM. Free commercial use below $5M in annual revenue.