1. GPT-3.5-turbo
2. GPT-4

GPT-3.5-turbo

• All plans are supported.
• OpenAI has updated the 3.5-turbo model to support 16k tokens by default, so GPT-3.5-turbo and GPT-3.5-turbo-16k are now merged.
• Deduction calculation: {Sent tokens / 8192 + Received tokens / 4096} rounded up. Approximately 2 tokens per Chinese character.
• Deduction order: pay-per-call usage > monthly subscription usage > perpetual usage.
• This rule also applies to API access on this platform. Please refer to the desktop settings -> API access for more information.

GPT-4

• Only perpetual usage plans support GPT-4.
• Due to the high price of GPT-4 and the affordability of monthly subscription plans, GPT-4 is not available for monthly subscriptions. Usage will be deducted based on actual usage, and a minimum of 10 remaining usages is required to use GPT-4.
• Approximately every 400 Tokens sent and every 200 Tokens received will be deducted. One English character is approximately 1 token, and one Chinese character is approximately 2 tokens. For more information on tokens, please refer to external resources.
• In summary, the price of GPT-4 may exceed your expectations.

Please note that this translation is for reference purposes only and may not be an exact representation of the original text.