Latest in AI
Every brief, newest first. Each one links to its primary source.
-
Moonshot’s Kimi K3 arrives on Amazon Bedrock
The open-weight 2.8-trillion-parameter model is now available behind AWS procurement terms, and is the first open model on Bedrock with explicit prompt caching.
-
xAI claims halved transcription errors at the same price
Grok Voice Transcribe 2.0 keeps version 1.0 pricing and includes speaker diarisation, though most of the accuracy evidence is xAI’s own.
-
Anthropic will relax Claude’s biology safeguards for verified labs
Verified life science teams will get access to Claude models with looser limits on biology work, monitored against a declared use case rather than…
-
Anthropic puts paid outside evaluators inside its own building
Anthropic will give evaluators from Accenture employee-level access to its model development, with no published standard yet for what they may report.
-
Cohere to absorb Aleph Alpha, with no price disclosed
Aleph Alpha, Germany’s best-known model developer, is merging into Cohere under the Cohere brand, with no price, valuation or ownership split disclosed.
-
Anthropic used Claude to speed up 30 open biology models
Anthropic reports roughly 4x average speed-ups after Claude spent four weeks optimising more than 30 open-source biology models. All figures are its own.
-
OpenAI launches Astra for Law, with its own benchmark numbers
OpenAI’s new legal configuration of GPT-6 Astra passed 54.0% of a research benchmark it ran itself, against 38.7% for the same model with plain…
-
Anthropic puts a number on how much AI research Claude leads
Anthropic says Claude now leads 26% of its own AI research work, up from under 1% in February, and wants rival labs to publish…
-
Google open-sources a way to fake the freight data nobody will share
MilleMiglia generates synthetic freight networks so researchers can benchmark middle-mile logistics without the proprietary data carriers keep private.
-
EU AI Board meets for the ninth time as transparency rules bed in
The EU body that coordinates national AI Act enforcement met on 17 September; the published readout lists priorities and dates but no decisions.