Traditional credit-based Enterprise agreements draw down workspace credits; Chat may be metered per message, while Work and Codex already use token-based credit metering. Under the token-based Enterprise model, Chat, Work, and Codex are instead billed in USD from actual token usage, with metered usage contributing to a monthly invoiced amount.

The current Standard rate for GPT-6.1 Sol in Work and Codex is $2 per 1M input tokens, $0.10 per 1M cached-input tokens, and $10 per 1M output tokens. A workload using 1M input + 1M cached input + 1M output therefore costs $12.10 in Standard mode. Under the current credit-based rate card, the same token mix would consume 302.5 credits (50 + 2.5 + 250). Those figures are not a fixed currency conversion because Enterprise agreements can include contract-specific discounts or other commercial terms.

The rate card also lists additional multipliers. Where applicable in Work and Codex, Fast mode is billed at 2× the Standard USD token rate and GPT-6 Astra Ultrafast at 6×. Long-context requests above 272K input tokens generally use 2× input and cached-input rates and 1.5× output rates, while GPT-6 Astra in Codex is explicitly exempt from the >272K long-context multiplier.

This billing model is available to eligible ChatGPT Enterprise agreements, including ChatGPT for Financial Services, and is not currently offered for ChatGPT Edu or ChatGPT for Healthcare. Existing credit-based workspaces do not switch automatically because of the rate-card update; OpenAI states that token-based rates apply only after an eligible new or replacement Enterprise agreement changes the workspace billing model.