AI

reflection/Beam-1.0-501B-A23B · Hugging Face

Read full article at source ↗

Based on a feed excerpt; full article content has not been acquired. Source evidence retrieved 17/10/2026, 00:47:26 UTC.

Briefings

Beam 1.0-501B-A23B, an open-source model developed by Reflection, merged llama.cpp support and introduced sparse MoE architecture with 501B total parameters and 23B active parameters. It supports 1M token context, BF16/FP8/NVFP4 numerics, and agentic coding with 43.3% SWE Bench Pro v2-Hard performance. The model was submitted by jacek2023 to r/LocalLLaMA.

Key points

• Beam 1.0-501B-A23B merged llama.cpp support and introduced sparse MoE architecture with 501B total parameters and 23B active parameters.
• Beam supports 1M token context, BF16/FP8/NVFP4 numerics, and agentic coding with 43.3% SWE Bench Pro v2-Hard performance.

Context

Beam is an open model built by Reflection, designed for agentic tool use, coding, and reasoning. It supports long-horizon task completion, MCP-based tool integration, and browsing, with graduate-level science problem solving capabilities.

Limitations

Evidence is limited to a single GitHub pull request and benchmark results from public splits; no additional background on model deployment or user adoption is supplied.

Generated by Bittrees-hosted qwen35-2b sha256:aaf42c8b7c3c on 25/10/2026, 05:24:20 UTC. Based on publisher-provided feed evidence; may not cover the full original article. Check the source for details.

View archived briefing on IPFS
bafybeiaiolkmjlz2yuwuxss3rkft3jpcjgdockp7pwxhoyaayiejywhsv4