Groq Api

[aborruso]

Hello and thank you for this wonderful project.

Is it possible to use also the groq API, which is openai compatible?

Thank you

[ncoop57]

This is now supported from main

[aborruso]

Thank you very much

[aborruso]

hi @ncoop57 how to set it?

I have set in this way my config file:

[DEFAULT]
provider = openai
model = llama-3.3-70b-versatile
base_url = https://api.groq.com/openai/v1
api_key = xxxxxx
history_lines = -1
code_theme = monokai
code_lexer = python

When I run ssage hi ShellSage I get

raise self._make_status_error_from_response(err.response) from None
openai.BadRequestError: Error code: 400 - {'error': {'message': "property 'max_completion_tokens' is unsupported, did you mean 'max_tokens'?", 'type': 'invalid_request_error'}}

Thank you

[ncoop57]

@aborruso

Weird... it looks like Groq does not follow the openai api spec properly: https://platform.openai.com/docs/api-reference/chat/create#chat-create-max_completion_tokens

Gonna go ahead and close this as this is a problem with Groq and not shell sage

[aborruso]

Thank you @ncoop57

Gonna go ahead and close this as this is a problem with Groq and not shell sage

I have seen that groq uses max_tokens and it is Deprecated :(