Stay informed. Keep your time.
8 stories · ~3 min readEstimated reading time for the entire edition.
Kimi K3.1 Pro’s weights were released under the Kimi K3.1 Pro license, and the model is also available through the Kimi platform API. The company says it offers a one-million-token context window and improved coding and 3D generation; evaluations used different tools and conditions.
Kimi K3.1 Pro has been released as an open-weight model and is available through an API.
Images 2.6 is available in ChatGPT and Codex; developers also get the Flare and Sunburst options for the API. According to the source, image generation is 50% faster than in the previous version. New tools include GIF creation, image marking in the API, and SVG export.
Images 2.6 generates images 50% faster than its predecessor and lets users create GIFs.
OpenAI announced it will acquire Pierre Computer Company, which develops the Trees and Diffs libraries used in Codex. After the acquisition closes, OpenAI aims to bring Code Storage features to Codex. It says open-source projects will remain open source and that the deal is subject to regulatory approval.
OpenAI announced it will acquire Pierre Computer Company to improve Codex’s Git and code-review tools.
FlashProve is an open-source inference-verification prototype that creates zkSNARK proofs linking responses to specific weights. In Anthropic’s 4-bit Llama 4 Maverick test, proof generation reached 110 tokens per second on one H100; a proof for a 1,024-token request was 18.78 MB, and generation used 70–95 times more GPU time than vLLM. The work is presented as interim progress, not a completed research program.
Anthropic released the open-source FlashProve prototype, designed to verify that a response came from a particular model without revealing its weights.
In Edgewater, Florida, voters added a ban to the city charter; in Perry, Ohio, they imposed financial and operational restrictions on hyperscalers seeking tax breaks. According to the report, financiers have begun treating local opposition as a credit risk.
Voters in more than 30 localities across nine states approved measures involving bans, moratoriums, or restrictions on data centers.
Linux can be installed on the MSI Prestige RTX Spark laptop, and CUDA acceleration works, but when plugged in, the GPU drops to about 780 MHz. In a GPT-OSS-120B test, Linux produced 13.9 tokens per second, versus 65.4 on Windows and 65.1 on WSL. The article attributes the issue to a Windows-specific power-management component with no Linux equivalent.
In tests on the MSI Prestige RTX Spark, its plugged-in Linux GPU stayed near 780 MHz, and GPT-OSS-120B ran slower than on Windows.
A researcher from Calif used AI tools during an OpenAI-sponsored PHP audit to turn a PHP flaw previously viewed as a memory bug into a vulnerability enabling server-side code execution. The flaw was reported to the PHP Foundation, and a team of four engineers prepared a patch in two months. The researchers said AI alone was not enough and expert knowledge also mattered.
AI-assisted review revealed that a PHP memory bug could be turned into a critical vulnerability allowing code execution on a server.
Geoffrey Hinton warned that the unmanageable risks of AI could cause events such as a major power outage or disruption to the financial system. Suggesting such an incident could happen this year or next, Hinton argues that the most advanced models should undergo mandatory, independent testing before release.
Hinton suggested a major AI-related outage could happen this year or next and called for independent pre-release testing of models.
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.
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.
