Daily AI Briefing
The 6 top stories of the 120 AIBriefs curated that day.
Get this briefing in your inbox every morning
Carbon tops CursorBench 4.0, Perplexity's WANDR and Zapier's AutomationBench at over 2X lower cost per task than the next best external model. Priced at $2/M input and $10/M output tokens, same as Gemini 4 Argon, which stays the pick for most software engineering work.
Fable 5.2 is up to 30% faster than Fable 5.1 and priced at $8/$40 per 1M tokens, down from $10/$50, with 1M context and $0.20/Mtok cache reads. Mythos 5.2 is pitched as the most capable model for cybersecurity and life sciences research, available through verification programs.
The offering covers 825,000,000 Class A shares, 801,351,354 from Anthropic and 23,648,646 from existing stockholders, with a 30-day underwriter option for 123,750,000 more. At the top of the range the listing implies a fully diluted valuation of $1.404 trillion, below earlier expectations of around $1.5 trillion.
Anthropic, PBC filed a Form S-1 for a Class A common stock offering and intends to list on the Nasdaq Global Select Market under the ticker "ANT"; share count and price range are not yet set. The 300-plus page prospectus reportedly discloses $417 billion in future AI spending.
Inkling 2 Small is a Mixture-of-Experts model with 594B total parameters and 17B active, scoring 31 on the Artificial Analysis Intelligence Index. It pretrained in 68 hours on NVIDIA GB300s with a native 4-bit NVFP4 forward pass, and is available on Tinker with Apache 2.0 weights planned next week.
METR's 11-week deep-access evaluation found Anthropic's models improving roughly twice as fast as a year ago, with ~1.9x acceleration since October 2025 and no clear bottlenecks. Anthropic redacted most of the evidence METR found most informative, citing risk of accelerating other developers.
Also that day
102 more stories that day — open the live feed.
Get tomorrow's AI brief in your inbox