OpenAI Unveils Powerful Unified AI Agent

OpenAI has released a new AI model, GPT-6, that combines chat, code generation, and web browsing capabilities into a single unified agent. The model has a 2M token context window and improved coding and reasoning abilities.

Details

OpenAI has released a new AI model, GPT-6, that combines the capabilities of ChatGPT, Codex, and web browsing into a single unified agent. The model has a 2M token context window, which allows it to handle large amounts of data, such as days of sensor data for IoT telemetry pipelines, without the need for chunking. The pricing for GPT-6 remains the same as GPT-5.4, but the model is 40% better at coding and reasoning. This represents a trend where model capability improves while cost stays flat, which the author sees as a subsidy war that benefits builders. The author believes that both unified agents and decomposed multi-agent systems will coexist, depending on the complexity of the problem. Unified agents will work well for straightforward workflows where a single model can handle the full context, while decomposed systems will still be useful for more complex domains where specialization is required.