OpenAI begins rolling out Astra model after warning of its advanced cyber capabilities
OpenAI said companies participating in its application-based cybersecurity program will be first to get access to Astra.
On Thursday, OpenAI unveiled Astra, its newest AI model and the company's most advanced yet. The tech giant claims Astra represents "a new frontier on computer and browser use," handling tasks with "speed, accuracy, and safety" unparalleled by any other model. OpenAI unveiled Astra to its cybersecurity program Daybreak customers on Thursday, and will make it available through the company's paid plans and API later this week.
In a press call, OpenAI president Greg Brockman stated Astra was the company's "most intelligent and, also very importantly, our most aligned model yet." He emphasized Astra's ability to "bring together years of our research and big bets, with each breakthrough having built on the last" and said it "represents a real shift in what kind of work people can delegate to AI and how it can empower them."
Astra's cyber capabilities are a major talking point. OpenAI published a blog earlier this week detailing the model's new abilities and new safety measures. The company tested Astra on various security benchmarks to ensure its capabilities, stating it can identify and develop zero-day exploits to help defenders find and patch weaknesses.
OpenAI also boasts Astra's coding abilities, claiming it's the "best model for software engineering to date." They provide benchmarking test results showing Astra outperforms other models, including OpenAI's own Sol and Anthropic's Fable, in tasks like finding bugs, executing terminal tasks, and answering codebase queries.
However, Astra's use of opaque recurrence, a reasoning technique that obscures chain of thought monitoring, adds controversy to the model. OpenAI downplays the extent of opaque recurrence, framing some opacity as a natural consequence of model evolution. Chief scientist Jakub Pachocki acknowledges monitoring a model's reasoning process becomes more challenging as capabilities increase.
A reporter asked if OpenAI is calling Astra the official arrival of AGI, to which Brockman clarified there is no longer a contractual AGI trigger, and AGI has evolved into a "mission concept or spiritual concept." He left the decision up to readers, but personally believes Astra signifies the arrival of AGI.
Written by urgent.news from TechCrunch's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
This story
This is one outlet's version. Read the fullest account.
- 'Welcome to the AGI era': OpenAI launches GPT-6 Astra venturebeat.com
- OpenAI launches GPT-6 Astra and says welcome to the “AGI era” thenewstack.io
- OpenAI releasing major upgrade to ChatGPT and Codex with GPT-6 Astra, details here 9to5mac.com
- OpenAI hails ‘new era of artificial general intelligence’ with Astra model release theguardian.com
- OpenAI launches Astra, its powerful (and controversial) new model techcrunch.com
- OpenAI launches new Astra model amid growing scrutiny over agents' safety channelnewsasia.com
- OpenAI launches GPT-6 Astra, initially available to organizations in its Daybreak Access program; Greg Brockman says "we are now in the AGI era" (Hayden Field/The Verge) theverge.com
- OpenAI says Astra was built on its largest-ever training run, using more than 100,000 GPUs at its Stargate site in Texas (Ina Fried/Axios) axios.com