# 11.4 Usage and limits (/managing-ai-workers/when-the-worker-goes-wrong/usage-and-limits) --- type: Document title: "11.4 Usage and limits" description: "How usage limits and a shared allowance can stop a recurring job, and how to plan and check usage on Claude and on ChatGPT." status: stable order: 211.4 ksor: owner: team:panaversity audience: [ public ] approval: by: process:panaversity at: 2026-10-08T13:21:13Z chapter: "11" part: II expert_status: required concepts: [ "11.4" ] last_verified: 2026-10-08 sources: - id: anthropic-cowork title: "Get started with Claude Cowork (Claude Help Center, updated 6 October 2026, verified 2026-10-08)" resource: https://support.claude.com/en/articles/13345190-get-started-with-claude-cowork - id: anthropic-usage-best-practices title: "Usage limit best practices (Claude Help Center, updated 22 September 2026, verified 2026-10-08)" resource: https://support.claude.com/en/articles/9797557-usage-limit-best-practices - id: anthropic-usage-limits title: "How do usage and length limits work? (Claude Help Center, updated 16 September 2026, verified 2026-10-08)" resource: https://support.claude.com/en/articles/11647753-how-do-usage-and-length-limits-work - id: anthropic-usage-credits title: "Manage usage credits for paid Claude plans (Claude Help Center, updated 7 October 2026, verified 2026-10-08)" resource: https://support.claude.com/en/articles/12429409-manage-usage-credits-for-paid-claude-plans - id: openai-codex-usage title: "Using Codex with your ChatGPT plan (OpenAI Help Center, verified 2026-10-08)" resource: https://help.openai.com/en/articles/11369540-using-codex-with-your-chatgpt-plan generated: at: 2026-10-08T13:21:13Z by: esl-rewrite/1.2.0+ksor.1 trust_tier: unverified build_id: sha256:2f1e14f2a0f11dbc37eb5856ca9b7d35b4ce4894da74c3328cf61a3c17c03861 dirty: true ksor_version: 0.0.60 --- **In everyday life.** A family shares one mobile data plan. One person streams videos all weekend, and by Monday the shared data is used up. The others did nothing wrong. **What you will learn.** How usage limits can stop a recurring job, and how to plan and check the budget each job spends from. You practice it in step 3 of the lab. **Why it matters.** Heavy work on one day can stop a recurring job the next day, and nobody is told. In this chapter's story, Maria is the office manager at Brightline Wholesale Supply, the book's example company. On Monday, January 11, 2027, she spent the afternoon on a large year-end task in the company's AI workspace. On Tuesday, the duplicate check that the company's AP Worker, an AI Worker, runs on each invoice email did not start. Each run said "Not started. Usage limit reached." The weekly limit came back on Wednesday morning. Agentic work costs more than chat.[^anthropic-cowork] A task that plans, searches and reads twenty files uses far more of your allowance than one answer does.[^openai-codex-usage] Your allowance is the amount of use your plan includes. Long conversations cost more too, because more context is read on every turn.[^anthropic-usage-limits] On plans with a fixed allowance, three facts are true on both AI vendors, Anthropic and OpenAI. First, work in different places in one account usually draws from **one pool**, which is a single shared allowance. Maria's year-end task used it up, so the duplicate check stopped. Second, limits are measured over two periods, a **five-hour window and a week**, so Monday's work can stop Tuesday's job. Third, **overflow is a cost decision**. Overflow is any use beyond your limit. At a limit you wait, upgrade or pay for more, and that is a budget choice, not a fix. So treat a limit as part of the design, not only as an error. *Check usage* before heavy work and before you blame a brief, the worker's written instructions. *Keep recurring work* on a seat, one person's account in the company's AI workspace, whose budget is planned for it, away from one-off work. *Spend less per run*: brief well the first time, because three rounds of corrections cost three runs. Never solve a limit by moving work to your personal account, outside the company's AI workspace. Work data in a tool or route the company never approved is shadow AI (Chapter 10). > **Anthropic, as verified 8 October 2026.** Fixed-allowance plans have a five-hour session limit and weekly limits, shown in *Settings > Usage*.[^anthropic-usage-best-practices] Claude, Claude Code and Claude Desktop share one limit.[^anthropic-usage-limits] At a limit you wait or upgrade.[^anthropic-usage-limits] On Pro and Max you can also buy usage credits, billed at API rates.[^anthropic-usage-credits] Usage-based Enterprise plans have no fixed limits.[^anthropic-usage-best-practices] There are more details in [The details on both AI vendors](both-ai-vendors.md). > **OpenAI, as verified 8 October 2026.** On plans with shared allowances, ChatGPT Work and Codex draw from one pool. The pool is measured in five-hour and weekly windows and shown in the usage dashboard. At a limit you add credits, use an available reset, upgrade or wait. OpenAI's list of what shares the pool does not include ordinary chat.[^openai-codex-usage] There are more details in [The details on both AI vendors](both-ai-vendors.md). **Comparison.** Both AI vendors have one pool, but it covers different work. On Anthropic, chat and coding share it. On OpenAI, ChatGPT Work and coding share it, and OpenAI's list leaves out chat. Before you choose where a recurring job runs, know what else uses up its pool. **What stays the same.** Every run spends from a budget someone owns. Plan the budget for recurring work. Check it before you look for the cause of a missing run. [^anthropic-cowork]: Get started with Claude Cowork, Claude Help Center. [^openai-codex-usage]: Using Codex with your ChatGPT plan, OpenAI Help Center. [^anthropic-usage-limits]: How do usage and length limits work?, Claude Help Center. [^anthropic-usage-best-practices]: Usage limit best practices, Claude Help Center. [^anthropic-usage-credits]: Manage usage credits for paid Claude plans, Claude Help Center.