Codex· OpenAIPolicy change
Codex moves to a token-based rate card
OpenAI began moving Codex pricing from average per-message estimates to token-based credit rates, with usage computed against input, cached input, and output tokens. The Codex pricing page continues to describe 5-hour usage windows and notes that additional weekly limits may apply.
At a glance
- Agent
- Codex · OpenAI
- Event type
- Policy change
- Announced
- April 2, 2026
- Effective
- April 2, 2026
- Plans affected
- ChatGPT PlusChatGPT ProBusinessEnterprise
What changed
- As of April 2, Codex rate cards started moving to credits per million input, cached input, and output tokens.
- Plus: GPT-5.5 shows 15–80 local messages per 5-hour window.
- Pro $100: 5× Plus by default, with a temporary 10× promo through May 31, 2026.
- Pro $200: 20× Plus by default, with a temporary 25× 5-hour limit promo through May 31, 2026.
- Enterprise / Edu flexible-pricing usage scales with credits rather than fixed per-seat limits.
Official record
This event is documented in OpenAI's own materials. The link below is the canonical source we cite.
Codex rate card (OpenAI Help Center)Additional references