Grok
Grok 4.7 arrives from SpaceXAI at the same price as the previous release
SpaceXAI released Grok 4.7 for coding, agentic tasks and knowledge work, keeping the $2/$6 per million token pricing of Grok 4.6.
Grok
SpaceXAI released Grok 4.7 for coding, agentic tasks and knowledge work, keeping the $2/$6 per million token pricing of Grok 4.6.
anthropic
Opus 5.5, the first model in the Claude 5.5 family, performs at the level of Fable 5.1 on most work, according to Anthropic. Token prices drop 20%, cache reads 60%, and output runs over 30% faster.
OpenAI
OpenAI has announced Sol and Luna, two GPT-6 models aimed at everyday work with different balances of capability and cost.
OpenAI
OpenAI has updated its smaller Sol and Luna models, offering API access at half the price of the previous 5.6 series.
Google says a Mandiant persona joined the inner circle of TeamPCP in March, watching the supply-chain hacking spree from the inside and helping revoke stolen credentials.
voice cloning
In Hume's voice replication leaderboard of 11 models, Fish Audio s2-pro scored 4.03 for speaker similarity while ElevenLabs Eleven v3 finished last at 2.91.
Claude Code
A quotation collected by Simon Willison describes a workplace where specs, code, tests, tickets and reports are all produced with Claude Code.
Amazon
Amazon has cut off Meta's Muse AI agent from buying on its platform, saying unauthorised agent access breaks its conditions of use.
StepFun
StepFun's new sparse MoE model holds about 600B total parameters, activates roughly 27B per token, and is set to get open weights on October 15, 2026.
Qwen
Qwen has released Qwen-Image-2.1, a 7B model with open weights that handles both image generation and editing in a single architecture.
China
The New York Times says China's artificial intelligence progress is in focus this week, while its economy is in its worst shape in decades.
OpenAI
A researcher documented how OpenAI's __obi cookie, set by bzr.openai.com, is sent back from advertiser websites and tied to a ChatGPT account.