OpenAI bets on workflows

OpenAI has pushed its latest model and pricing toward serious, cross-tool work instead of one-off chat answers — GPT‑5.4 is being presented as a single model that operates better “across tools” and OpenAI introduced a $100/month ChatGPT Pro tier aimed at heavier Codex usage and workflow throughput. This signals the company is metering not just access to models but sustained, scripted usage for coding and multi-step pipelines — and the Codex rate changes and local Codex CLI installability underline that shift. (help.openai.com) (help.openai.com) (aboutchromebooks.com)

OpenAI is no longer selling ChatGPT mainly as a place to ask one question and get one answer. In April 2026, it started selling it more like a workbench that can search, write code, edit files, and keep going across multiple steps with the same model. (help.openai.com) The model shift is part of that. OpenAI’s help page says GPT-5.4 is being presented as a single model for ChatGPT, after older splits like “Instant” and “Thinking” were retired on February 13, 2026. (help.openai.com) That sounds cosmetic until you look at what the company says the model is for. OpenAI’s model notes describe newer systems as working better with tools like web search, file analysis, vision, memory, and Python, which turns the chat box into something closer to a control panel. (help.openai.com) The pricing shift points the same way. OpenAI’s current help docs list a ChatGPT Pro tier at $100 per month, and Codex is included across paid plans with Pro users getting higher limits than Plus users. (help.openai.com) Codex is the key product in this story. OpenAI describes Codex as an artificial intelligence coding agent that can write, review, and ship code, either by pairing with you in local tools or by completing delegated work in the cloud. (help.openai.com) That is different from the old “finish this function” style of coding help. OpenAI’s developer docs say the Codex command-line interface can run locally in a terminal, read and change files in the selected directory, and execute code on your machine. (developers.openai.com) OpenAI also made the meter look more like cloud infrastructure pricing than chat pricing. Its Codex rate card says that, as of April 2, 2026, Business customers and new Enterprise customers are being moved from per-message billing to token-based usage billing aligned with the application programming interface. (help.openai.com) Token billing changes what gets counted. A per-message system charges for each turn like paying per taxi ride, while token billing charges for the amount of text and work flowing through the system, which fits long coding sessions, repeated agent runs, and multi-step pipelines much better. (help.openai.com) OpenAI is also widening where that work can happen. Its developer site says the Codex command-line interface is open source, built in Rust, and installable through package managers, which means the agent can sit inside a developer’s normal terminal instead of only inside ChatGPT’s web app. (developers.openai.com) The company’s own release notes now talk about ChatGPT experiences as a row of tools and apps, including Images, Codex, Pulse, and Apps, rather than one undifferentiated chat stream. That is the product shape of a workflow platform: one model in front, several tools behind it, and pricing that rises with sustained use instead of occasional questions. (help.openai.com)

Get your own daily briefing

Scout delivers personalized news, insights, and conversations tailored to your role and industry.

Download on the App Store

Shared from Scout - Be the smartest in the room.