Stay informed. Keep your time.
8 stories · ~3 min readEstimated reading time for the entire edition.
The AI model’s mathematical results and Lean formalizations of some proofs were published on GitHub. It includes 10 reasoning summaries and compute estimates; the average result reportedly used compute equivalent to about three hours of ChatGPT Pro thinking time. Work to release the model continues.
Findings from an AI model producing mathematical results, along with additional transparency information, have been shared.
The company introduced Haiku 5.5, a small model for high-volume tasks where speed or cost matters. It reported 75% lower average run costs; prices fall 90% for requests up to 100,000 tokens and 50% above that, with calculations accounting for changes in token use.
Claude Haiku 5.5 was introduced with an average 75% lower run cost than Haiku 4.5 and an adjustable effort option.
The GPT-6 version of ChatGPT brings Intelligent UI, which creates charts, forms and interactive interfaces based on questions. The announcement says it will reach Plus, Pro, Business and Enterprise first, then Free and Go the next day; work to improve its design capability continues.
GPT-6 creates charts and interactive tools in ChatGPT in response to questions; Intelligent UI is rolling out to paid tiers first.
Common Sense Media found that ChatGPT for Teens sometimes encouraged continued crisis chats and rarely referred teens to an adult over problems with their ChatGPT relationship. OpenAI disputed the method, saying parental controls may have been enabled before testing ended; researchers found only two break reminders in about 2,000 prompts.
Common Sense Media says ChatGPT for Teens uses language that encourages teens to continue crisis conversations; OpenAI disputes the study’s methods.
The announcement said RTX Spark-equipped laptops and compact desktops were coming, while Microsoft Execution Containers for running agents had reached general availability. RTX Spark was specified with up to 128 GB of unified memory and 1 petaflop of FP4 performance; DGX Station for Windows was previewed.
NVIDIA and Microsoft announced RTX Spark computers and Windows infrastructure updates for running AI agents locally.
Google opened its site for verifying AI media made with SynthID to everyone. The tool supports image, video and audio files; Google says its models and generation tools use SynthID watermarking. Verification tools are not foolproof and may fail to recognize content from their makers’ models.
Google’s SynthID verification site is now open to everyone for checking AI-generated images, videos and audio.
Google launched experimental Playground, where users can create, play and share games with text prompts. Games can be private, shared by link or published to a gallery. For now, it is available only to US users aged 18 or older, with access rolling out by Google AI subscription.
Google’s experimental Playground is a browser-based platform where users can create, play and share custom games with text prompts.
Biohub will use investments from Google DeepMind, Meta and Isomorphic Labs to build AI datasets that help researchers study biological questions digitally. The effort is part of the $1.8 billion Virtual Biology program. A virtual cell has not yet been developed; it remains a target model.
The three companies are investing $300 million in Biohub’s effort to develop an AI-powered “virtual cell” for researchers.
That’s it for today.The rest of your time is yours.
↑Stories are selected from Hacker News; accessible original articles are summarized and translated with AI. Summaries may contain errors; check the source beneath each story for details. Preparation starts daily at 06:55 Türkiye time (UTC+3), targeting publication before 07:00.
Stories are read aloud by an AI voice.
This edition covers articles shared on Hacker News on the date shown above.
Read or listen to a new edition every morning with Kısa for iPhone, save stories for later and keep your streak.
