Purchased 1000 credits, gone in under an hour

OpenAI changed the pricing and usage in ChatGPT for Codex credits - still with an unclear and unpublished amount that you get included with a subscription. It is now a direct consumption from your balance based on token usage, at the same pricing you’d get if using the API.

With: the disadvantage that you are forced to buy large increments, only useful in ChatGPT for Codex, virtually ensuring breakage (purchased but unused credits kept by OpenAI). The mapping of “credits” to the value of them is also highly-obfuscated.

Updated 3 days ago

Codex usage is priced based on API token usage, calculated as credits per million input tokens, cached input tokens and output tokens. Learn more about tokens here.

This format replaces average per-message estimates with a direct mapping between token usage and credits. It is most useful when you want a clearer view of how input, cached input, and output affect credit consumption.

Under this model, actual credit usage depends on the mix of input, cached input, and output tokens in each task. The table below displays credits per 1M tokens for each token type.