Platform hub

Codex limits

Local message allowances, five-hour windows, credits, and plan-dependent agent usage. Values below are scoped rather than flattened into one misleading platform-wide number.

Verified Aug 22, 2026Official docs
Quick answer:  Codex has multiple independent constraints. Match the exact plan, model, runtime, invocation mode, or server configuration shown in each row.

Current documented limits

ConstraintCurrent valueScopeVerified source
Plus local-message rangeThe official Plus column ranges from 5–45 for GPT-6 Astra and 10–100 for GPT-5.6 Sol to 250–2,000 for GPT-5.6 Luna.5–2,000 messages per 5 hours, depending on modelChatGPT PlusOpenAISep 17, 2026
Pro 5× local-message rangeThe Pro 5× column ranges from 25–225 for GPT-6 Astra and 50–500 for GPT-5.6 Sol to 1,250–10,000 for GPT-5.6 Luna.25–10,000 messages per 5 hours, depending on modelChatGPT Pro 5×OpenAISep 17, 2026
Pro 20× local-message rangeThe Pro 20× column ranges from 100–900 for GPT-6 Astra and 200–2,000 for GPT-5.6 Sol to 5,000–40,000 for GPT-5.6 Luna.100–40,000 messages per 5 hours, depending on modelChatGPT Pro 20×OpenAISep 17, 2026
post-allowance continuationPlus and Pro can purchase credits, while API-key chats use standard API billing.Additional credits or API-key usagePlan dependentOpenAISep 17, 2026

How to apply Codex limits safely

The monitored baseline covers Plus local-message range, Pro 5× local-message range, Pro 20× local-message range, post-allowance continuation. Treat these as separate constraints rather than one platform-wide capacity number: a workload can fit one row and still fail another because the plan, model, endpoint, runtime, region, invocation mode, or account scope differs.

  1. Match the production workload to the exact scope printed beside each value and confirm it in the active Codex console, configuration, or response headers.
  2. Measure the serialized request, token volume, duration, concurrency, storage, or connection demand at realistic percentiles, then preserve headroom for bursts and retries.
  3. Check every adjacent layer—client, SDK, gateway, proxy, queue, database, and downstream service—for a smaller effective limit before changing architecture.

Specific limit pages

These pages exist because the constraint has a distinct implementation or troubleshooting intent. Closely related keyword variations stay consolidated.

Compare alternatives

Official sources

View Codex change history