OpenAI pricing & models
What happened
OpenAI has started treating model usage as a cost-and-workflow problem by publishing a Codex rate card that differentiates coding credits across consumer and enterprise plans. Reports also say OpenAI rolled out GPT-5.4 with a 1‑million token context window and native desktop features, and that a new base model nicknamed “Spud” is being positioned as a long‑term ChatGPT foundation for future releases (help.openai.com, wwwhatsnew.com, storyboard18.com).
Why it matters
OpenAI published an updated Codex rate card on April 2, 2026 that converts many coding credits to a per-token billing model and lists different credit rates for ChatGPT consumer plans and for Business and Enterprise plans. (help.openai.com) OpenAI released a new model called GPT-5.4 on March 5, 2026; the company ships it inside ChatGPT and in paid developer products, and it adds built-in ability for the model to perform actions on a desktop or in web apps while keeping about one million units of text in short-term memory. (openai.com) A “token” is a unit of text roughly equivalent to a few characters or a single short word, and switching to per-token billing charges customers for the actual text the model processes (reads and writes) instead of charging per chat message; OpenAI’s help page and developer rate card say credits will be consumed based on which model is used, the size and complexity of the task, and whether the task runs locally or in the cloud. (help.openai.com) (developers.openai.com) “Native computer use” means the model can interact with application user interfaces—clicking, typing, opening files, and invoking menus—without extra wrapper scripts, enabling end-to-end automated workflows, and the million-token context window is the amount of text the model can consider at once, which lets it load very large codebases, entire long documents, or multi-file project histories into a single session. (openai.com) (datacamp.com) The Codex help entry also notes short-term promotional changes: for a limited time Codex access is included in the ChatGPT Free and Go tiers and other plans have temporary increased rate limits, while the new token-aligned rate card is already applicable to ChatGPT Business and Enterprise customers and will be rolled out for other plan types in coming weeks per the developer guidance. (help.openai.com) (developers.openai.com) OpenAI is concurrently preparing a new foundational model nicknamed “Spud,” which company leadership has described as a long-term base for future ChatGPT releases and which external reporting says has been under development for roughly two years; that planning has been reported alongside internal shifts in executive responsibilities to prioritize infrastructure and scaling. (msn.com) (theinformation.com)
Key numbers
- OpenAI published an updated Codex rate card on April 2, 2026 that converts many coding credits to a per-token billing model and lists different credit rates for ChatGPT consumer plans and for Business and Enterprise plans.
What happens next
- OpenAI published an updated Codex rate card on April 2, 2026 that converts many coding credits to a per-token billing model and lists different credit rates for ChatGPT consumer plans and for Business and Enterprise plans.
- (msn.com) (theinformation.com) OpenAI has started treating model usage as a cost-and-workflow problem by publishing a Codex rate card that differentiates coding credits across consumer and enterprise plans.
Quick answers
What happened in OpenAI pricing & models?
OpenAI has started treating model usage as a cost-and-workflow problem by publishing a Codex rate card that differentiates coding credits across consumer and enterprise plans. Reports also say OpenAI rolled out GPT-5.4 with a 1‑million token context window and native desktop features, and that a new base model nicknamed “Spud” is being positioned as a long‑term ChatGPT foundation for future releases (help.openai.com, wwwhatsnew.com, storyboard18.com).
Why does OpenAI pricing & models matter?
OpenAI published an updated Codex rate card on April 2, 2026 that converts many coding credits to a per-token billing model and lists different credit rates for ChatGPT consumer plans and for Business and Enterprise plans. (help.openai.com) OpenAI released a new model called GPT-5.4 on March 5, 2026; the company ships it inside ChatGPT and in paid developer products, and it adds built-in ability for the model to perform actions on a desktop or in web apps while keeping about one million units of text in short-term memory. (openai.com) A “token” is a unit of text roughly equivalent to a few characters or a single short word, and switching to per-token billing charges customers for the actual text the model processes (reads and writes) instead of charging per chat message; OpenAI’s help page and developer rate card say credits will be consumed based on which model is used, the size and complexity of the task, and whether the task runs locally or in the cloud. (help.openai.com) (developers.openai.com) “Native computer use” means the model can interact with application user interfaces—clicking, typing, opening files, and invoking menus—without extra wrapper scripts, enabling end-to-end automated workflows, and the million-token context window is the amount of text the model can consider at once, which lets it load very large codebases, entire long documents, or multi-file project histories into a single session. (openai.com) (datacamp.com) The Codex help entry also notes short-term promotional changes: for a limited time Codex access is included in the ChatGPT Free and Go tiers and other plans have temporary increased rate limits, while the new token-aligned rate card is already applicable to ChatGPT Business and Enterprise customers and will be rolled out for other plan types in coming weeks per the developer guidance. (help.openai.com) (developers.openai.com) OpenAI is concurrently preparing a new foundational model nicknamed “Spud,” which company leadership has described as a long-term base for future ChatGPT releases and which external reporting says has been under development for roughly two years; that planning has been reported alongside internal shifts in executive responsibilities to prioritize infrastructure and scaling. (msn.com) (theinformation.com)
