AI Summary
5 min readOpenAI Launches GPT-6 Astra, Calling It the Start of AGI
OpenAI has released GPT-6 Astra, which the company is branding as the beginning of artificial general intelligence. The model was trained on over a thousand GPUs at the Texas Stargate site, and Greg Brockman called it "a generational leap towards artificial general intelligence." On benchmarks, GPT-6 scored 98.6% on Arc AGI 3 and a perfect 100% on Exploit Bench, meaning it is particularly strong at hacking, coding, science, and computer use — the ability to operate a user's computer directly.
The model is rolling out first to organizations with "daybreak access," with subscribers to follow. Sam Altman posted that "we hope it will begin to enable a new generation of entrepreneurship, scientific discovery and building" and called it "the best model in the world for computer use, professional work, science, coding, cybersecurity, and more." He also acknowledged that extra time was needed to meet safety and alignment standards.
Continue reading the full summary in the app — free to try.
Read Full Summary →Free • No credit card required
Never miss an episode of Hard Fork AI
Get every new episode summarized in your inbox — free, ~5 minutes to read.
No spam. Unsubscribe anytime.
What you'll learn
- 1 (00:01) **Episode Overview & Top Stories** - Host introduces the major AI news items: OpenAI’s GPT-6 Astra release, a simultaneous multi-model outage, OpenAI walking away from a Cursor deal, Meta’s Muse Spark discount, and NVIDIA acquiring Hugging Face.
- 2 (01:21) **Sponsorship: AIBox Platform** - Host promotes AIBox.ai, a platform with 90+ top AI models (audio, video, image, text) that allows users to compare responses from different models in a single chat thread.
- 3 (02:20) **OpenAI’s GPT-6 Astra Release** - The host details the specifications and benchmarks of the new model, trained on over 1,000 GPUs at the Texas Stargate site, with Brockman calling it a "generational leap towards AGI."
- 4 (04:04) **Sam Altman’s Statement & Demo Highlights** - The host reads Sam Altman’s tweet claiming the model is the best for computer use, professional work, science, and cybersecurity, noting it took extra time for safety and alignment.
- 5 (07:46) **Simultaneous Outage: ChatGPT, Claude, and Grok** - The host transitions to the story of all three major models going down within 90 minutes of each other, suggesting a potential "domino effect" from the GPT-6 launch.
- 6 (09:43) **Infrastructure Concerns & SpaceX’s Role** - The host analyzes the implications of the outage, arguing it proves the risk of putting "all of our eggs in one infrastructural basket" (SpaceX).
- 7 (10:44) **OpenAI Walks Away from Billion-Dollar Cursor Deal** - The host explains that OpenAI ended its partnership with Cursor after SpaceX acquired the company, a deal projected to bring over a billion dollars annually.
+ Full timestamped outline available in the app
Show Notes
In this episode, we present a hardcore analysis of the service failures that occurred during the launch of OpenAI's GPT-6, especially affecting ChatGPT and Claude. We explore future implications.
Show Links
More from this podcast
Hard Fork AI →