🤖OpenAI Unveils GPT-6 Astra: AGI for Real-World Tasks
Astra can now do your job, across all apps
TL;DR
OpenAI's GPT-6 Astra is a new AGI model designed to navigate software like a human, performing tasks across browsers, spreadsheets, and more. It scored 72.6% on OSWorld 2.0 and 97.6% on ARC-AGI-3, marking a significant leap in AI training.
OpenAI just dropped GPT-6 Astra, an AGI model designed to handle real-world tasks across various applications. Astra can fill out forms, update CRM records, and even create 3D games. It scored 72.6% on OSWorld 2.0 and 97.6% on ARC-AGI-3, showcasing its versatility. If you're automating workflows or building AI-driven tools, this is a big deal. Astra's training used over 100,000 DBUs, setting a new standard for model complexity.
Key Points
GPT-6 Astra scored 72.6% on OSWorld 2.0, taking 40 minutes per task.
Astra achieved 97.6% on ARC-AGI-3, setting a new benchmark for AGI.
The model scored 95.9% on BenchCAD and 96% on GPQA Diamond.
Astra's training used over 100,000 DBUs, the most ever for an OpenAI model.
NVIDIA's AVO architecture scored 100% on ARC-AGI-3 with a 30% baseline.
Why It Matters
If you're automating workflows or building AI-driven tools, GPT-6 Astra is a must-watch. It can handle complex tasks across applications, from filling out forms to creating 3D games. The model's performance on benchmarks like OSWorld 2.0 and ARC-AGI-3 signals a significant leap in AI capabilities, making it a game changer for developers and businesses alike.
Frequently Asked Questions
Why does this matter?
If you're automating workflows or building AI-driven tools, GPT-6 Astra is a must-watch. It can handle complex tasks across applications, from filling out forms to creating 3D games. The model's performance on benchmarks like OSWorld 2.0 and ARC-AGI-3 signals a significant leap in AI capabilities, making it a game changer for developers and businesses alike.
What happened?
OpenAI's GPT-6 Astra is a new AGI model designed to navigate software like a human, performing tasks across browsers, spreadsheets, and more. It scored 72.6% on OSWorld 2.0 and 97.6% on ARC-AGI-3, marking a significant leap in AI training.
Comments
Be the first to comment
Enjoyed this article?
Get it daily. 7am. Free. Reads in 5 minutes.
Join 3,465 builders reading daily.