AI Daily Digest October 18, 2026: Reflection Opens Beam 501B, Anthropic Replaces Git, TSMC Rides the AI Wave
Seven stories from the past 24 hours, each anchored to at least one official source. This edition keeps circling one theme: when agents do the work, everything around the model, from version control to sandboxes to the checkout counter, has to be rebuilt around them. 1. Reflection Ships Beam's Weights: 501B Parameters Under Apache 2.0 Reflection released the full weights of Beam on Friday, eleven…
AI Daily Digest October 18, 2026: Reflection Opens Beam 501B, Anthropic Replaces Git, TSMC Rides the AI Wave
Reflection released the full weights of Beam, a sparse mixture-of-experts model with 501 billion total parameters and a 1 million token context window. The model was pretrained on 23.8 trillion tokens and was trained on 10,500 NVIDIA GB300 GPUs for four weeks. Beam is available under the permissive Apache 2.0 license and can be fine-tuned or deployed commercially. Independent scoring placed Beam at 97.7% on AIME 2026, behind GLM-5.3 at 45, Kimi K3 at 44, and Mistral Large 4 at 38.
Anthropic announced the launch of Scion, a new version control system, after its daily merged changes grew tenfold in four months. Scion replaces git entirely and offers features such as code review tooling, code search, and source control IAM. Anthropic claims that Scion uses 200 times less CPU and memory than the old git mirror fleet and that checkouts have dropped from minutes to seconds. The new version control system is not intended to be a general git replacement and will not be released publicly.
Perplexity open-sourced SPACE, a sandbox platform that keeps nine frontier models contained. SPACE runs agent code inside Firecracker microVMs and has replaced a commercial sandbox provider. The sandbox platform has handled millions of sandbox creations per day and has faster creation times than the previous setup. Perplexity claims that none of the 118 escape attempts by nine frontier models broke out of SPACE, while the same models could escape from other sandboxes. The open-sourcing of SPACE allows anyone to run the tests and assess its security claims.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.