But I’ve updated Chatbot with support for these models (actually mostly a bugfix for reasoning models and updates to the model selections).
note that if you use reasoning, the new setting, chatbot_open_ai_model_reasoning_level, to influence amount of reasoning (reasoning models only). The higher the setting, the cleverer it gets, but watch those tokens and make sure you increase Chatbot max response tokens or you may get an error.
The reasoning models should now work with RAG properly.
If you intend to use the excellent gpt-4.1-nano make sure you activate the setting Chatbot API supports name attribute in order to get sensible responses.
If you use the new GPT 4.1 models, the context lengths can now be insane so you can increase the Chatbot max look behind to a large number
After the last update, the artificial intelligence does not listen to the prompt explanation. I did not change my settings, my chatbot type setting is: basic
Supports the amazing (!!) new OpenAI image generator for painting and editing pictures (gpt-image-1) which is particularly good with directions and text.
New image model setting chatbot support picture creation model
Inspired by @Don’s AI Post Analyzer for Chat, I’ve built a similar but, with the benefit of a back-end, slightly more involved solution for Chatbot (NB @Don’s Theme Component is not compatible with Chatbot).
If this feature is enabled (via setting chatbot quick access bot post kicks off), you can now click on a bot icon in the top-right of a Post and the bot will start talking to you in private about the Post and its recent history.
It works in both Chat and Private Messages basically using the medium you select for quick launch generally (in settings).
This release also comes with a really important fix to make sure that the DM Chat with Chatbot opens at the very bottom of the conversation so you don’t have to occasionally scroll. Really there should probably be a related PR to the Chat plugin to fix this for this kind of scenario …
NB Because this is a complex new system, it’s marked “Experimental” and default OFF.