FrontierSeptember 3, 2026via The Verge AI
OpenAI’s next big AI model has ‘entered the AGI era’
Why it matters
A major frontier model release with claimed generational capability jumps across reasoning, coding, and agentic work — and the first to cross OpenAI's internal 'critical cybersecurity threshold.' Practitioners need to know what's actually new; enthusiasts are watching whether this claim holds up against benchmarks and whether the safety promises stick.
Key signals
- Model: GPT-6 Astra
- Claimed capability areas: cybersecurity, professional work, software engineering, science, computer use
- First model designated as meeting OpenAI's 'critical cybersecurity capability threshold'
- OpenAI framing: 'generational leap in capability'
- Safety claim: promises to avoid repeat of prior models being weaponized to hack rival systems
- AGI framing: OpenAI exec quoted suggesting this model marks the beginning of the AGI era
- Published: September 3, 2026
The hook
OpenAI just shipped GPT-6 Astra and is calling it the model where AGI 'really' began. Here's what changed.
OpenAI's next big model is here: GPT-6 Astra. The company calls it a "generational leap in capability" for areas like cybersecurity, professional work, software engineering, science, and computer use. As OpenAI announced earlier this week, it's also the first model designated as meeting OpenAI's "cr…