The blog posts collectively highlight significant advancements in ChatGPT Enterprise, Edu, and Codex usage management, focusing on flexible and automated budgeting tools, robust API enhancements, and scalable deployment strategies. The primary trends and announcements include:

  • Introduction of flexible, fine-grained budget release schedules (hourly, daily, weekly) for ChatGPT usage, with capabilities to adjust allocations for high-demand users and optimize resource availability.
  • Enhanced ChatGPT Admin API features, enabling comprehensive management of workspace identities, usage limits, credit and USD billing, and user-specific overrides—all supported by strong authentication and requirement validation procedures.
  • Deployment frameworks for automated budget control leveraging AWS services (Lambda, EventBridge, SQS, DynamoDB, Secrets Manager), emphasizing secure, reliable, and scalable task orchestration.
  • Step-by-step pilots and starter kits (with Node.js examples and Codex integration) that assist organizations in simulating, setting up, and transitioning to managed and automated infrastructure for budgeting processes.
  • Detailed guidance on configuration, cohort management, policy design (fixed, individual, headroom-based), and compliance with workspace billing—alongside best practices for credential security, enrollment snapshotting, and membership renewal.
  • Comprehensive testing and validation disciplines, covering both local and cloud environments, multi-platform compatibility, and rigorous documentation to ensure reliable, auditable budget management deployments.

Together, these updates empower organizations to efficiently automate and tailor ChatGPT usage limits, ensure regulatory and billing compliance, and streamline deployment and testing of usage-budget management systems.

New Cookbook Recipes

README.md

Source: openai/openai-cookbook

The blog post announces the introduction of flexible ChatGPT usage budgets for Enterprise and Edu users, allowing them to automate updates to usage limits via the ChatGPT Admin API. Users can now release portions of their monthly budget—up to 2,000 credits—on an hourly, daily, or weekly basis, accommodating ongoing needs throughout the month. This feature includes the ability to adjust budget amounts based on user consumption, ensuring that high-demand users can access more resources as needed. A runnable Node.js example is provided, enabling users to customize and test their release plans. Guidance is offered for implementation, including setup for various environments, testing strategies, and managing budget allocations effectively. Overall, these enhancements aim to optimize resource availability and improve user experience in managing ChatGPT usage.

api-contract.md

Source: openai/openai-cookbook

The blog post outlines the requirements and features of the ChatGPT Admin API at https://api.chatgpt.com/v1. It emphasizes the necessity of using a workspace-scoped ChatGPT Admin key for authentication and details the specific operations permitted under various key permissions, such as managing workspace identities, usage limits, and user overrides.

Key features include a connection check via AWS to verify secrets and access before member capture or budget configuration. The adapter manages credit and USD units for usage settings, requires careful handling of overrides and historical data during restoration, and stresses the importance of verifying settings before live use. It also covers the renewal of membership periods, ensuring that all members have a settled state and that settings align with current standings. The blog post serves as a comprehensive guide for using the API effectively and responsibly.

aws.md

Source: openai/openai-cookbook

The blog post outlines the process for deploying a serverless controller on AWS to manage scheduled releases using AWS Lambda, EventBridge Scheduler, and SQS. Key features include a framework that enhances scalability and efficiency in task processing while maintaining data integrity with DynamoDB and Secrets Manager for sensitive information.

The deployment process is initiated with a disabled configuration to allow for testing and tuning, emphasizing a methodical approach to ensure reliability and performance. The post details preparation steps, testing protocols, and environment setup requirements, including the need for specific AWS services, configurations, and permissions.

It further discusses methods to verify AWS delivery and workspace connectivity, ensuring proper access to necessary resources. Finally, the article presents steps for preparing and previewing budgets, which involves snapshotting enrollment data and preparing controls for live operation while maintaining strict security practices regarding sensitive information.

codex.md

Source: openai/openai-cookbook

The blog post announces a usage-budget pilot with Codex, detailing a step-by-step setup and execution process. Users can begin by requesting Codex setup and running a demonstration that simulates fictional user scenarios on various platforms. Key features include executing commands for enrollment, budget limits, and restoration procedures using Node.js. The post emphasizes rehearsals for both credit and dollar-based budgets, outlining commands to inspect, approve, and run simulations. Additionally, it discusses reviewing small pilots, managing scheduled tasks, and verifying the clean-up of pilot components. Finally, it instructs users to transition to a managed infrastructure, ensuring a smooth continuation of their Codex operations with provisioned environments and scheduled renewals. This pilot is aimed at organizations looking to effectively manage usage limits and automate processes within Codex.

get-started.md

Source: openai/openai-cookbook

The blog post introduces a starter kit for implementing a budget management system for ChatGPT usage. It provides a structured approach for organizations to set up weekly release plans using fictional scenarios depicting credit and dollar billing units. Users can download the starter ZIP file containing essential files and explore budget amounts and schedules via an HTML interface.

Key features include running examples with Node.js (version 24 or later), simulating releases that adjust monthly budgets, and using Codex for assistance. The article outlines steps to run the demo, set up managed environments, and establish pilot programs tailored to specific user groups. Additionally, it emphasizes the importance of using cloud services for ongoing management and renewal processes to accommodate budget adjustments for future periods.

local.md

Source: openai/openai-cookbook

The blog post outlines the procedures for implementing and testing a controller on macOS or Linux for managing workspace budget limits using Node.js. It details the setup for rehearsing workflows, including initializing directories, generating snapshot enrollments, and applying configurations for both credit and dollar units. Users are advised to prepare fictional members for simulations and utilize local schedulers for testing without a model. It includes specific commands for rehearsals, generating and inspecting scheduler files, and creating live service setups. Emphasis is placed on securely managing credentials, generating appropriate service files, and ensuring that users thoroughly review configurations before executing live changes. Additionally, it provides distinct processes for macOS and Linux, highlighting credential storage and system service management for effective deployment.

operations.md

Source: openai/openai-cookbook

The blog post provides a comprehensive guide on configuring and operating a usage-budget controller for managing individuals’ spending limits in a workspace environment. Key features include selecting billing units (credits or USD) and defining budget release policies such as fixed budget release, individual starting limits, and observed usage headroom. The guide outlines necessary steps for setting up a budget, configuring cohorts of users, capturing current usage data, and preparing a reviewed enrollment with specific user IDs or emails. Instructions for pausing, restoring, and renewing the budget controller are also detailed. The focus is on ensuring compliance with workspace billing agreements and maintaining accurate usage tracking while implementing tailored budget plans. Admins are advised to keep their configurations secure and orderly throughout the process.

verification.md

Source: openai/openai-cookbook

The blog post outlines a comprehensive approach to testing and validating a usage-budget example in a multi-platform environment using Node.js and AWS. It covers 32 combinations of policy and unit intervals through simulated APIs without making actual requests. Key features include automated tests for policy handling, debt management demonstrations in both credits and USD, and AWS runtime checks to ensure reliability.

Users are guided on executing specific commands to run tests, verify source integration, and prepare AWS packages. The post emphasizes thorough documentation and evidence collection for live deployments, stressing the completion of acceptance checks and how to handle periodic renewals. It underscores the importance of maintaining accurate records of source commits and test results throughout the process.