-
industry · OpenAI · 2건
OpenAI models are coming to Google Cloud
Google Cloud will offer GPT 6.1 Astra, Sol, and Luna in the coming weeks, according to Ars Technica. OpenAI says it and Google Cloud are bringing frontier models and Codex workflows to Model Garden, available through Google Cloud Marketplace.
Teams that already buy and run through Google Cloud Marketplace and Model Garden can evaluate OpenAI models and Codex workflows in the same environment, possibly without a separate vendor contract.
-
research · Hacker News
PHP2Shell: AI uncovers critical RCE in PHP
A Hacker News post titled PHP2Shell reports that AI uncovered a remote code execution (RCE) vulnerability in PHP. The title describes the flaw as critical.
Teams running PHP in production should check for patch advisories on this RCE and review their update schedule.
-
product · Hacker News
Using Claude to reduce the attack surface of code
A post titled "Using Claude to reduce the attack surface of code" appeared on Hacker News. Its title indicates it covers using Claude to shrink the attack surface of code.
Developers weighing Claude for security reviews or refactoring can look at it as an example of using the model to cut a codebase's attack surface.
-
industry · Hacker News
Anthropic logged net loss of over $80B in first half of 2026
A Hacker News post reports that Anthropic recorded a net loss of over $80B in the first half of 2026. No further details beyond the headline figure are given.
Developers building on the Anthropic API may want to check provider-switching options and cost caps, since a loss of this size could eventually bear on pricing.
-
research · Hacker News
Claude designed protein binders against scorpion neurotoxins
A Hacker News post reports that Claude designed protein binders targeting scorpion neurotoxins. The title points to a case of Claude being used for protein design work.
Developers wiring Claude into scientific workflows like protein design need to decide in their pipeline where the model's design step ends and wet-lab validation begins.
-
research · Hugging Face
NVIDIA Kumo Relational Outperforms Supervised Machine Learning on Relational Data, No Training Required
A Hugging Face post says NVIDIA Kumo Relational outperforms supervised machine learning on relational data. According to the title, it requires no training.
Teams running prediction on relational data can now weigh a training-free approach against maintaining their own supervised training pipelines.