OpenAI to merge ChatGPT Chat and Work tabs into a single experience

Merging Chat and Work

OpenAI has announced plans to test a unified experience within ChatGPT that brings the Chat and Work tabs together. This move follows the launch of ChatGPT Work just four months ago, which initially offered a separate space for agentic tasks. The company intends to start the test with a small group of users before a wider rollout.

The goal of this merger is to allow users to think through an idea and have it carried out immediately in one place. Users will be able to provide files and plugins to ask the AI to build websites, create presentations, or complete tasks on the web. This removes the need for users to manually choose between a standard chatbot and an agentic tool.

This strategy mirrors a move recently made by Anthropic with its Claude product. Anthropic merged its Chat and Cowork tabs so that the model determines if a prompt requires a fast response or a complex sequence of actions involving research and code. OpenAI is following a similar path to simplify the user experience by letting the model handle the task selection.

Unified usage allowances

A significant part of the overhaul involves how OpenAI will track and limit user activity. ChatGPT and Codex will move to a single shared allowance instead of running on separate counters. Previously, users had distinct limits for Chat, Work, and Codex, along with separate limits for images, voice, and file uploads.

OpenAI stated that it is aiming for a smooth transition and believes the majority of customers will not find their limits resetting more frequently than they currently do. However, consumption will depend on the specific model and the complexity of the task. More advanced models and complex agentic tasks will cost more against the shared pool of credits.

To help manage this transition, OpenAI recently increased the Work and Codex limits for its Pro tiers, which range from $100 to $500 per month. Users will be able to monitor their remaining total allowance and see when it resets under the Settings and Usage menu. Those who run out of credits will have the option to wait, upgrade their plan, or purchase additional credits.

The cost of agentic tasks

The move toward a shared allowance is driven by the high computational costs associated with agentic behavior. Agents that work for hours, click through browsers, or build websites consume significantly more compute than a standard chat response. Industry estimates suggest that tasks for some agents can use five to twenty times more compute than a simple chat.

By merging the pools, OpenAI shifts the decision of how to spend compute from the company to the user. This makes it easier for OpenAI to price its services because separate pools for chat, agents, and coding are difficult to price cleanly. It also addresses the fact that Codex is used by more than five million people a week and has become a significant cost factor for the company.

For casual users, the impact may be minimal because the 'Instant' fast default model remains unlimited and does not count toward the shared allowance. However, for heavy users who subscribed to ChatGPT Work specifically for its ability to perform long-running tasks with powerful models, the shared limit creates a new trade-off. They will now have to balance their high-intensity work tasks against their general chat usage.

User reaction and next steps

The announcement has already sparked debate among subscribers who may feel they are losing a benefit. Some users on social media have expressed concerns that they effectively had two separate allowances for the price of one subscription. These users have voiced concerns about how exactly usage will be counted and some have even threatened to cancel their subscriptions.

OpenAI has a history of adjusting limits to manage demand, having added up to 50 percent more usage in August after some initial issues with the Work and Codex allowances. The company is now looking for feedback from its test group to refine the experience before the full rollout. They aim to ensure that the transition to a unified experience is smooth for the majority of the user base.

Moving forward, the success of this merger will depend on how well the model handles the distinction between quick answers and multi-step jobs. While Anthropic has already faced criticism from heavy users who find that coding sessions drain their chat limits, OpenAI will face similar scrutiny. The company will need to balance the technical costs of agentic work with the needs of its power users.