[FREE PLUGIN] AI Proxy - ChatGPT, OpenAI, Claude, LlaMA, PaLM Real-Time Streaming

Ill test both. If you need context, you need to take the messages currently and merge them into the json.

I will see if I can add a way to ‘preserve context’ or fix the issue you are talking about

1 Like

That would be fantastic! This plugin is already far superior to the others i tried until I came across your version. They are all clunky and lacking in comparison. One of them has Stream’s Message History state which was handy but unfortunately was not as flexible and pretty slow in comparison. Thanks again for your time and hard work on bringing this game changing plugin to the community!

Sorry to bust balls, I have been trying merge messages into JSON, kinda goes back to what I was asking @web4 in how he went about doing this, what is the best practice for doing this. Anyone who can help me with this I will be forever grateful. Last piece of the puzzle for me. Thanks in advance.

Would love to know this also. I am having the same problem as above with message history and with Add Messages, System. How to go about the “merge” into JSON? an “AProxyStream A’s Message History” would be a godsend.

It’s OK I found a workaround by adding message history into the JSON’s system message. How would that work out token wise?

1 Like

Tokens are all those that are send

I scheduled this for this weekend, if I dont get the flu.

2 Likes

@gaimed I’m having some issues saving data to the database in the workflows… have you ran into any issues saving items to the database? Can you make a video on this?

For added context, When I open debug mode / inspector, it shows "make changes to thing… “AIproxystreamA’s messages:firstitem: (Empty)” but the same logic on my UI displays the proper message that I’m trying to save with the same logic.

The way I save to database is adding to workflow, when AIproxystream A Is generation finisihed > Create new thing ,works fine for me.

Solved! Thanks for the heads up!

Hi @gaimed :wave: Does the streaming work with Azure?

yes, custom endpoint

Hi @gaimed , for OpenRouter, if the model I want to use is not listed in the list. can I still use custom endpoint to open router to achieve this? Is there anything to lookout for ?

image

yes, just specify the custom model name

Awesome plugin, I like the combination of openrouter and openai. Your website seems to be offline: https://nocodesaastoolbox.com/

Can you please update me on this one.
Thanks

Hi there, I disabled new signups for a couple of days. Will enable once there is more capacity.

Hey Ab, do you have an estimated timeframe for when signups will be re-enabled?

Its live again on https://aiproxy.stream/

1 Like

Thank you!!

Hi @gaimed - great app. One thing we noticed - our responses are getting cut off/not completing all the way, any way we can customize the max_tokens or temperature variables? We are using OpenAI’s GPT-4 model.

We are trying to add the JSON params at the AI token stage but that does not seem to work.