• New method authorize_cost: used to ensure the user is willing to spend enough points before you make potentially expensive model calls
• New method capture_cost: used to capture the actual cost after you perform the model calls
• New setting cost_label: used to show a short pricing string in the Poe UI (e.g. "100+ points")
• New setting rate_card (previously custom_rate_card): used to provide a more detailed explanation of your pricing structure

Add get_bot_response_sync, which is a synchronous wrapper around get_bot_response. This allows users to call the function more easily without having to worry about asyncio.

Fix an issue with release 0.0.57 where attachments were not made if post_message_attachment was called at the end of get_reponse.

1. Bots can now pass attachments to each other via fp.stream_request. To do this, call the post_message_attachment function, which continues to work like before, but will now also implicitly emit a file event to other server bots calling it. Calling bots will receive the file in the attachment field of a PartialResponse from the bot.
2. OpenAI tool calling improvement: the request will finish without making an extra unnecessary second call to the OpenAI bot if no tools are decided to be called. This saves cost to the user.

This release adds the new DataResponse object which bots can yield to attach a metadata string to their response. This string can be accessed in the metadata field of the associated ProtocolMessage in future requests to the bot. This is useful for storing additional information about previous messages in the conversation.

If allow_attachments and expand_text_attachments are both True, PDF files uploaded by the users will have their contents automatically appended to the conversation for the model to reference. This previously only worked for text and html files.

Previously, when defining server_bot_dependencies there was a limit of 10 calls to other bots for a single message. This limit has been increased to 100 calls for a single message.

• Updated the suggested optimized Poe Previews prompt to better support React. See https://creator.poe.com/docs/how-to-create-a-prompt-bot#optimize-prompt-for-previews for the full prompt

• We updated the fastapi_poe library to require Python >=3.9, as Python 3.8 reached end of life in October 2024.
• This version also includes a fix for tool calls failing when the model decides not to call the function.

• Added ability for server bots to help users create prompt bots with pre-filled fields. See https://creator.poe.com/docs/server-bots-functional-guides#creating-bots-that-help-their-users-create-prompt-bots for more details.