Just now, Altman announced on
For developers, especially "shrimp farmers", the main benefit is not only simplicity and convenience, but also the overall lower cost of tokens.
In the past, using the API to run a task of "monitoring emails around the clock and replying" might generate hundreds of dollars in bills every month; now, chatgpt Plus or Pro level subscribers no longer need to purchase a large number of additional API Tokens. They only need to use the original subscription to call GPT models of the same level in OpenClaw, including GPT-5.5's Terminal-Bench 2.0 (command line workflow understanding), GDPval (knowledge work index) and other core intelligent agent model capabilities.
OpenClaw was founded at the end of 2025. It is an open source AI agent project created by the well-known independent developer Peter Steinberger. It focuses on "being able to complete various tasks." It can connect to various mainstream large models and let AI take over users' daily tasks through WeChat groups, Slack, iMessage and other instant messaging applications or computer instructions - processing mailboxes, writing code, booking flights, operating hardware and even running banking interface operations.
Interestingly,In early April this year, OpenAI’s biggest competitor, Anthropic (Company A) founded by former OpenAI members, suddenly announced the opposite strategy to OpenAI: banning OpenClaw.Starting April 4, 2026, subscribers will no longer be able to use Claude subscription credits to access their large models through third-party integration tools such as OpenClaw. If the user wants to continue to let OpenClaw call Claude, they must pay on a pay-as-you-go basis using the regular API and pay a new fee that is billed separately from the original subscription plan. This action has brought a lot of trouble to a large number of user groups who rely on open source AI agents to complete automated work.
OpenClaw founder Peter Steinberger, who has joined OpenAI, fiercely criticized Anthropic's move at the time: "The timing is too coincidental - they first copied some popular features into their own closed-source integration tools (referring to Anthropic's own Claude Cowork and other intelligent agent products), and then blocked open source projects."
This wave of OpenAI has gained favor from a large number of developers. A developer left a message under Ultraman
Why did OpenAI become Open all of a sudden? The fundamental reason lies in the change of model thinking.When GPT-5.5 was released not long ago, the official repeatedly emphasized the highly "intelligent" characteristics of the model: "Our most intelligent, most intuitive and easy-to-use model to date", and also "an engine built for real work and agent tasks." Ultraman rarely made a high-profile statement in multiple reviews, "Based on my experience, it knows what to do." GPT-5.5 will regain its share in the fierce market with competitors such as Claude.
In other words, OpenAI's model is no longer focused on question and answer, search, drawing or writing copywriting, but on completing tasks - of course, this is also the direction that leading domestic open source models such as Qianwen and Kimi are working hard.ChatGPT's support for direct login to OpenClaw also means that in the Agent era, AI subscription services are no longer "users must come to my platform to use it", but "users can take AI with them after subscribing": an AI service system that can inside everything, drive everything, and integrate into everything. After logging in with an account, users can use AI subscriptions in any agent, workflow, or application scenario.
The opposite strategies of OpenAI and Anhropic towards OpenClaw reflect that the competitive logic of AI giants is undergoing a fundamental change: OpenAI is gradually becoming "Open", embracing the open source co-construction ecosystem, and realizing collective AI innovation; Anthropic is moving towards absolute closure, pursuing self-operated ecology and closed-loop security.