Beam: Reflection's 501B open-weight model
- ID
- 32087
- Status
- summarized
- Published
- 06 Oct 2026, 3:16 AM
- Fetched
- 06 Oct 2026, 7:01 AM
- Provider
- Hacker News
- Category
- dev-community
- Original URL
- https://reflection.ai/blog/introducing-beam
- Source URL
- https://hnrss.org/best
Summary
- Score
- 6.0
- Created
- 06 Oct 2026, 7:02 AM
- Tags
- Audience
- developersai_ml_learnersai_agent_users
What happened
Reflection announced Beam, its first open-weight model: a sparse Mixture-of-Experts with 501B total parameters and 23B active, aimed at coding, reasoning, and agentic workloads. It was pretrained on 23.8T tokens and went through an RL run of over 100M rollouts on 10.5K NVIDIA GB300 GPUs over 4 weeks. Weights, technical report, model card, and developer artifacts are promised later this month, with early access signup open; benchmarks claim competitiveness with GLM 5.2 and approach to Qwen 3.8-Max, plus inference efficiency.
Why it matters
No immediate action: the model is not released, and the benchmarks are vendor-reported with some baselines missing. Once weights, license, and model card are out, evaluate Beam for coding/agent tasks if you can handle a 501B-total MoE, where the 23B-active design may help serving cost but likely still needs serious hardware. Wait for independent evals and quantization/serving support before changing your stack.
Discussion angle
When the weights and technical report land, what would you actually test before switching a coding/agent workload: the vendor-reported SWE Bench/Terminal Bench deltas, the license, tool-calling reliability, or the cost of serving 501B total parameters with 23B active?