Reflection's Beam lands on Hugging Face: 501B MoE, 1M context, Apache 2.0 open weights
tomaarsen · x · 2026-10-16
Reflection's first model Beam is now on Hugging Face under the Apache 2.0 license.
Key specs:
- MoE with 501B total parameters, 23B active per token; 52 layers, 256 experts per layer with 8 routed
- Text-in/text-out with a 1M-token context window
- Checkpoints in BF16, FP8, and NVFP4
- Five reasoning levels (low/medium/high/xhigh/max), in-context reasoning persistence (preservethinking), and tool calling with a dedicated grammar
Beam targets coding and agentic workloads for teams wanting frontier-scale models on their own hardware, and ships with an open-source Rust-based harness plus a DFlash2 speculative decoding drafter.
Day-0 support covers transformers, vLLM, and SGLang; full-parameter RL is possible via Miles, and the post includes SFT and RL recipes.
More from Models
- Golden Gate Claude explainer resurfaces: interpretability premise still holds — HarperSCarroll · 2026-10-16
- Hugging Face livestreams release of a new open-weight model — huggingface · 2026-10-16
- Step 5 Preview hits #2 on OpenRouter weekly leaderboard with 19.5T tokens in one week — StepFun_ai · 2026-10-16
- Meta's open Muse Spark needs 493GB of memory — far beyond any consumer laptop — TejasKumar_ · 2026-10-16
- Apache 2.0 is the underrated part: Beam beats rival open drop on license — bigblueboo · 2026-10-16
- Apollo Research: GLM-5.3 spontaneously reasons about how its actions shape its future training data — Confident_Salt_8108 · 2026-10-16