
GPT-6 In 7 Minutes
GPT-6 (GPT Astra) Released: Benchmarks, Pricing, Computer Use Demos & What’s New OpenAI has released GPT-6, also called GPT Astra, and the video gives an overview of its capabilities, benchmarks, pricing, and examples, emphasizing state-of-the-art computer and browser use for real-world workflows. Astra is positioned as highly intelligent and aligned, with highlighted results on Agents Last Exam, Automation Bench, Screenshot Pro, and near-saturation on Arc AGI3 at 99.9%, plus strong performance on scientific workflows versus Fable 5.1. Pricing is listed at $10/M input tokens, $50/M output tokens, $1/M cached inputs, and $12.50/M cache writes, with a reported cost per task around 1.67. New features include asynchronous tool calling, midturn steering, and adjustable reasoning effort while preserving cache; it supports up to 1M context tokens and 128k output tokens, image input but not audio/video input, and uses text output with real-time voice in the ChatGPT desktop app. Examples include a detailed Manhattan world built in Unreal Engine and other immersive 3D environments, while safety notes include a 0% score on Exploit Gym; the video closes by previewing an upcoming tutorial on generating 3D assets and scenes. 00:00 GPT-6 Launch Overview 00:20 Computer Use Highlights 00:58 Benchmarks and Alignment 02:04 Pricing and Value 03:07 New API Features 03:25 Context and Multimodal 03:50 Showcase 3D Examples 04:30 Safety and Blog Notes 04:57 Why Computer Agents Matter 05:50 Speed Gains and Use Cases 06:54 Wrap Up and Next Video
Technical content at the intersection of AI and development. Building with AI agents, Claude Code, and modern dev tools - then showing you exactly how it works.
Weekly deep dives on AI agents, coding tools, and building with LLMs - delivered to your inbox.
Free forever. No spam.
Subscribe Free
New tutorials, open-source projects, and deep dives on coding agents - delivered weekly.