John RodriguesAI Design Engineer | Designing 0→1 AI native Products

OpenAI Launches GPT-6 Astra: 72% Accuracy Benchmark. #openai #astra

4 Sept 2026 · 1 min

Watch on YouTube

0:46

In this brief YouTube commentary, John Rodrigues announces OpenAI's release of the GPT-6 Astra model, highlighting its 72% benchmark accuracy, which he equates to human-level performance at that rate. He interprets the milestone as a signal that OpenAI is moving closer to artificial general intelligence (AGI) and transitioning from purely chat-based interactions to more agent-oriented models. Rodrigues emphasizes that the model's capabilities make it a valuable tool for developers looking to accelerate AI-driven workflows, and he invites viewers to share their own AI projects in the comments.

Chapters

There's a new update from OpenAI. OpenAI launched this new model called GPT-6 Astra. It's one of their most powerful models. It hit the benchmark of 72%. What that means is it's equally accurate as a human with 72% of accuracy to do the computer. This is why OpenAI is hinting towards more towards AGI and saying that we are going from a chat to be the agents and models being more worth. 72% is definitely a big benchmark to compare to all the models they have previously. And even if you are like building with AI, this model is going to be a really good one for you to accelerate some of your workflows and stuff. Yeah, worth looking into Astra and let me know what you are building with AI in the comments.