Beam 1.0 (Beam-1.0-501B-A23B) is Reflection's open-weight sparse mixture-of-experts reasoning model for coding and long-horizon agentic workflows. 501.3B total parameters with 23.4B active per token, 52 layers, 256 experts with 8 routed per token, text in and text out, 1,048,576 token context. Released under Apache 2.0 on 2026-10-16. Weights on Hugging Face as BF16, FP8 and NVFP4. Day-0 support in transformers, vLLM and SGLang.
No benchmark data available yet.
- Typetext
- Context1.0M tokens (~524 books)
- ReleasedOct 2026
- LicenseOpen Source
- StatusActive
- Cost / Message~$0.002
Frequently Asked Questions
Related Models
Qwen3 Coder Next FP8 · AlibabaQwen2.5 Coder 32B Instruct AWQ · AlibabaSpeaker Diarization Community 1 · PyannoteJais 3 0.9B · MBZUAIGemma 3 12B (free) · Google DeepMindBenchmarks
Key facts · as of 2026-10-16
- Beam 1.0 by Reflection. Not enough public benchmark scores to rank yet.
- List price $0.25 input · $1.00 output per 1M tokens (as of 2026-10-16).
- Sold by 1 provider (as of 2026-10-16): Nebius (fp8) $0.25 in / $1.00 out.
How to cite · data as of 2026-10-16
Beam 1.0 · benchmarks, pricing and providers. BenchGecko, data as of 2026-10-16. https://benchgecko.ai/model/beam-1-0-501b-a23b
Credit "Source: BenchGecko" with a link. Prices per provider and Gecko Tests are BenchGecko data (CC BY 4.0); benchmark scores keep their original source, listed in the JSON. JSON · llms.txt · MCP