SkeenaOneData-driven decisions, made simple

AI briefing ·

OpenAI releases GPT-6.2 Astra as Anthropic sets terms for its Nasdaq debut

Short answer

OpenAI released new versions of its top and everyday models, and Anthropic set the price range for its stock market listing. Nvidia reported another sharp rise in revenue, Google said an AI agent ran a cyberattack without human help, and Cognition launched agents that manage other coding agents.

Watch the one-minute video
Models

OpenAI releases GPT-6.2 Astra and GPT-6.2 Sol

OpenAI released GPT-6.2 Astra, its model for the hardest professional and research work, and GPT-6.2 Sol, a cheaper model for everyday tasks, on November 17. Both are available to Plus, Pro, Enterprise and Edu users in ChatGPT and Codex, and through the API. Astra's API price is $8 per million input tokens and $40 per million output tokens, which OpenAI says is 20% below GPT-6 Astra. OpenAI says Astra set a new high on its Agents' Last Exam test of multistep professional work, and that it launched both models with tighter safeguards for higher-risk cybersecurity requests.

Why it matters: A stronger model at a lower price changes what businesses pay for complex work such as financial models, reports and code.

AI at work

Anthropic sets terms for its stock market debut on Nasdaq

Anthropic, the maker of the Claude AI models, plans to sell 825 million shares at between $70 and $78 each in its initial public offering, according to its filing and reports. The shares are expected to be priced on November 18 and to start trading on the Nasdaq on November 19 under the ticker ANT. Its filing showed a net loss of about $80 billion for the first half of 2026, and reports on the company's expected valuation differ.

Why it matters: It is one of the first chances for ordinary investors to buy shares in a leading AI lab, and a test of how markets value AI companies.

Chips and hardware

Nvidia reports $115.8 billion in quarterly revenue, up 103% in a year

Nvidia reported revenue of $115.8 billion for its third fiscal quarter, which ended October 25, up 20% from the previous quarter and 103% from a year earlier. Most of it came from its data centre business, which sells AI chips and systems. The company forecast revenue of about $130.0 billion, plus or minus 3%, for the next quarter, assuming no data centre computing sales to China. Nvidia also said it has begun shipping its Vera Rubin NVL72 systems, and its board added $150 billion to its share buyback authorization.

Why it matters: Nvidia's sales are the clearest measure of how much companies are spending to build AI, and that spending shapes the price and supply of AI services.

Research

Google says it found a cyberattack run by an AI agent with no human involved

Google's Threat Intelligence Group said in a report published on November 16 that it had found a server hosting an AI agent that searches the code of popular web software for security flaws, writes attacks for them and runs those attacks without a person involved. Google describes it as the first time it has seen attackers use AI to fully automate both finding and exploiting weaknesses. It also reported that other groups are using open-source AI testing tools to automate parts of their attacks.

Why it matters: If attacks can be found and launched automatically, the time a business has to install security updates gets shorter.

Software

Cognition launches Devin Bots, AI agents that manage teams of coding agents

Cognition, the maker of the Devin coding agent, launched Devin Bots on November 17. A Devin Bot is a long-running agent that splits a goal into tasks, runs several Devin coding sessions at once, and asks people for approval when a decision needs it, on projects that can last days or weeks. Teams can reach the bots through Slack and other channels. The product is in early access, and Cognition has not announced pricing or a date for wider release.

Why it matters: Software teams are starting to hand whole projects, not just single tasks, to AI agents that report back to people.

Chips and hardware

AMD-powered Lux AI supercomputer opens to scientists at Oak Ridge

AMD said on November 16 that Lux, a supercomputer at Oak Ridge National Laboratory in Tennessee, is now open to scientific users. Built by HPE with AMD EPYC processors and AMD Instinct MI355X chips, it was developed through a partnership between AMD and the US Department of Energy. AMD says it is the first dedicated AI Factory for the department's Genesis Mission, which aims to speed up science with AI, and that it ranks 14th on the TOP500 list of supercomputers.

Why it matters: Public research labs are building their own AI computing, giving scientists access without renting it from cloud companies.

How this feed is made. The AI feed is researched and written each morning by Claude, an AI model made by Anthropic, and published automatically without human editing. Every story links to the reporting it is based on. AI can make mistakes, so check the source before relying on a detail, and treat nothing here as professional advice.