Add support for image inputs in ChatGroq

[AndRossi]

Checked

• I searched existing ideas and did not find a similar one
• I added a very descriptive title
• I've clearly described the feature request and motivation for it

Feature request

Langchain offers a smooth interface to connect with Groq APIs via package langchain-groq and its object ChatGroq. The interface is analogous to what you offer for other LLM providers, e.g., OpenAI or Anthropic.

Unfortunately, as reported in the official documentation, unlike ChatOpenAI and ChatAnthropic, ChatGroq does not support multimodal inputs.

This is a huge limitation: several models available on Groq support images, audios, and even videos.

Motivation

I would like to try Groq models for my multimodal RAG application, but I cannot as long as image inputs are not supported.

This is a bit frustrating - Groq is a major provider of open-source LLMs, and it is especially useful when users need lightning-fast inference, which is my case.

Proposal (If applicable)

No response