Published Frontier reporting tagged “models”.
10 articles · Full archive · Topics
5 recent pieces
Named partners, preclinical metrics, Claude Science, and a rare-disease programme that may reach human trials in late 2028 if candidates hold.
Read the piece5 stories
A terabyte of Apple unified memory is suddenly part of Meta’s open-weight pitch, three days after Wonder hit a public API price list.
NaiveAI says the 309B mixture-of-experts model has 15.5B active parameters and a native 1M-token context with no full-attention layers. It lists API prices of $0.10 input and $0.40 output per million tokens but gives no launch date.
Anthropic says Sonnet 5.5 runs more than 30% faster and costs up to 30% less per task. Anthropic also calls it the first Sonnet with cyber safeguards and classifiers that prevent reasoning extraction.
Mid-tier GPT-6 models under Astra. Sol $2 input and $10 output per MTok, Luna $0.10 and $0.50; Free and Go get Luna on desktop only; not yet in Chat.
First Claude release since the lab's pace-the-frontier call. API id claude-opus-5-5; $4/$20 per MTok and $0.20 cache reads; AWS, GCP, and Azure named.
New releases, then the constraint.
