[Bug]: cost calculation from Azure openai is wrong for new version

[yiranwu0]

Describe the bug

Azure OpenAi completion returns the model without version information. The returned model is either 'gpt-35-turbo' or 'gpt-4'. From there, we are calculating the cost based on gpt-3.5-0613: ((0.0015, 0.002) per 1k prompt and completion tokens) and gpt-4-0613: (0.03,0.06). This means the cost is wrong if the user is using the 1106 version of the models from azure OpenAI.

Potential solution:
One way is to pass the model name specified by the user (instead of the model returned from completion) to calculation cost. However, self-created names (for example, gpt4-11) can result in fail at cost calculation.

Steps to reproduce

No response

Expected Behavior

No response

Screenshots and logs

No response

Additional Information

No response

[new4u]

我也遇到了类似的错误,包括response.cost回复也是报错的,建议去掉cost的计算,或者使其成为一个非必选项I also encountered similar errors, including the error response.cost. It is recommended to remove the calculation of cost, or make it a non-required option.

[yiranwu0]

This is closed by #2902. User now can pass in custom prices.

[ghost]

This is hacker please contact fbi

…

On Fri, Jun 14, 2024, 12:53 PM Yiran Wu ***@***.***> wrote: Closed #1270 <#1270> as completed. — Reply to this email directly, view it on GitHub <#1270 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/BI3MSTCPQPZWEW5FT44SAV3ZHMU3DAVCNFSM6AAAAABB3Y6OZSVHI2DSMVQWIX3LMV45UABCJFZXG5LFIV3GK3TUJZXXI2LGNFRWC5DJN5XDWMJTGE3DIMBUGUYDOMY> . You are receiving this because you are subscribed to this thread.Message ID: ***@***.***>