Cognition Supports ChatGPT Personal Subscription Payment for Devin's GPT Usage
Cognition announced that users can connect their personal ChatGPT Go, Plus, or Pro plans to Devin, with all GPT model usage deducted from the ChatGPT plan's quota.
After connecting, enabling token sharing allows requests made by Devin on behalf of users to be deducted from the ChatGPT plan usage, without counting against Devin's own quota or on-demand limits. This feature covers Devin Cloud, Devin Desktop, and Devin CLI, with a single switch controlling synchronization across all three platforms. Users can complete account linking and permission authorization by selecting Continue with ChatGPT in the settings or switching in the model selector.
The OpenAI models constitute one of the model combinations in Devin Cloud's Lite, Normal, Ultra, and Fusion modes. After connecting, the payment for the OpenAI model portion in ChatGPT plan sessions will be deducted from the ChatGPT plan, while other models will still be deducted from the Devin quota. Users can allow Devin to use the entire ChatGPT quota or set a lower limit specifically for Devin in the settings. Combining Astra from the ChatGPT plan with SWE-2 from the Devin plan can make quota usage more efficient.
This feature is part of OpenAI's Sign in with ChatGPT plan sharing mechanism, which allows third-party applications to use the OpenAI model usage from users' existing subscriptions without needing API keys or separate billing. The heads of OpenAI ChatGPT and Codex stated that this mechanism has covered over 60 partner products, including Devin, OpenCode, and Lovable. Cognition quickly integrated after the mechanism went live, covering all its product forms.
The Devin Fusion architecture can use strong models for planning and review roles, while lower-cost models serve execution roles, matching cutting-edge model performance on coding agent benchmarks while reducing costs. After connecting to ChatGPT, users can incorporate usage of cutting-edge OpenAI models into their existing subscriptions, using them in parallel with Devin's own model quotas. Documentation confirms that Go, Plus, and Pro plans are all eligible, and sharing occurs across products after a single link.
Market Mechanism: OpenAI shares subscription quotas across applications, locking consumer-side payments to its own model calls, with funds flowing from third-party applications' independent pay-per-use billing to OpenAI subscription renewals. Coding agents like Devin benefit from lower user entry barriers and increased model usage stickiness, while independent API call model suppliers face pressure. Users can extend their existing subscription quotas to more tools, while enterprises face pressure to redistribute quotas across multiple applications.
Source: Public Information
ABAB AI Insight
Cognition was founded by Scott Wu, Walden Yan, and others, launching Devin in 2024 as the first end-to-end AI software engineer, emphasizing autonomous planning, coding, testing, and fixing. The company subsequently introduced the SWE-2 dedicated coding model and Fusion multi-model orchestration, combining strong model planning with low-cost model execution, driving 88% of merged pull requests in internal testing. It has established a multi-year strategic partnership with AWS to help clients modernize legacy systems and partnered with MongoDB to accelerate application migration to Atlas.
On the capital path, Cognition positions Devin as a deployable autonomous engineer, selling through the AWS Marketplace and collaborating with enterprise clients like Mercedes-Benz to reduce COBOL modernization cycles from months to days. After integrating OpenAI subscription sharing, the company reduces its reliance on its own quota sales, instead leveraging users' existing ChatGPT paid quotas to drive OpenAI model calls while maintaining differentiation with its own models like SWE-2.
This is similar to early SaaS platforms allowing users to bring their own cloud accounts or API keys, or payment infrastructures like Stripe opening billing layers to third-party applications. The current coding agent market is in an expansion phase from independent subscriptions to cross-platform quota sharing, with Devin reducing customer acquisition friction and expanding model usage by integrating mainstream consumer subscriptions.
Essentially, this represents a transfer of pricing power. OpenAI shifts the payment relationship for model calls from application-side pay-per-use to consumer subscription quotas, making third-party tools the entry point for consuming subscription quotas. Application providers lose some independent pricing space and instead compete on experience and efficiency within quotas; OpenAI, on the other hand, increases subscription retention and usage depth through multi-application consumption, concentrating capital from tool-level billing to subscription locking for model providers.
ABAB News · Cognitive Law
When subscription quotas are shared across applications, model providers lock in payment relationships.
Tools with independent billing that access shared quotas see pricing power shift upstream.
Users' existing subscriptions become entry subsidies for new tools.