Reflection’s new open‑weight model for coding and long‑term AI workflows is now listed on OpenRouter.
NoteBy OossaPublished by Oossa: 1 min read
OpenRouter now lists Reflection’s Beam 1.0 model. The model is built for reasoning tasks like code generation and multi‑step agent workflows. It uses a sparse mixture‑of‑experts design with 501 billion total parameters but only 23 billion active per token. The context window can hold up to 1,048,576 tokens.
Developers can run larger, more capable reasoning models without paying for full 501 billion‑parameter costs.
Tools & agents
Replit’s new Monitoring Agent can detect app outages, create a fix and deploy it, giving solo developers an on‑call engineer for their apps.
Tools & agents
Morgan Stanley, JPMorgan, Bank of America, Wells Fargo and others are stopping AI tools that log in with user credentials.
Tools & agents
Agent Compute Engine already powers Gemini and is expected to enter preview in the coming months. Google says it can pause idle agents to save capacity.
Oossa · Newsletter
The week in AI, explained
Every Monday: the stories worth knowing, in plain language. Free, no spam.
