signalHacker News Best2026-10-06
Beam: Reflection's 501B open-weight model
Reflection introduces Beam, a sparse Mixture-of-Experts model with 501 billion total parameters and 23 billion active. It was pretrained on 23.8 trillion tokens, and a high-compute RL run used 10.5K NVIDIA GB300 GPUs for four weeks, generating over 100 million rollouts. Beam matches GLM 5.2 on reasoning with 3-4x less inference compute and approaches Qwen 3.8-Max on coding and agentic tasks.
- for who
- Software engineers and enterprises building coding, reasoning, and agentic AI applications.
- why now
- Early access open for Reflection's new 501B open-weight Beam, weights due this month.
- what changes
- Enterprises can deploy a frontier-level open model at reduced inference cost, making heavy coding and agentic workloads more practical.
- to do
- Sign up for early access and prepare to integrate Beam's weights and developer artifacts when released later this month.
key points
- Beam is a sparse Mixture-of-Experts model with 501B total and 23B active parameters
- Pretrained on 23.8 trillion tokens, RL run used 10.5K GB300 GPUs over 4 weeks
- Matches GLM 5.2 on reasoning with 3-4x less inference compute, approaches Qwen 3.8-Max on coding
#open-weight model#MoE#reinforcement learning#coding agent
score
score 8 out of 10. 0-10: how dense the facts are, multiplied by how much you can do with them after reading. 8+ means the topic's evidence bar is met: benchmarks and availability for a new model, amount and investors for a funding round, revenue figures for a solo-money story. Below 5 an item does not enter the digest. A press release scores 3 or less, a reprint loses 2, anything older than 14 days loses 1, a headline that misleads loses 3.
read the source