Feature Request: Support Ollama Models

[xxlv]

Summary

Currently, ADK Go only supports Gemini models through the model/gemini package. This issue proposes adding support for Ollama, an open-source platform that allows running large language models locally.

Motivation

Supporting Ollama would provide several benefits:

1. Local Development: Enable developers to run agents locally without requiring API keys or internet connectivity
2. Cost Efficiency: Use open-source models without API costs
3. Privacy: Process sensitive data locally without sending it to external services
4. Model Variety: Access a wide range of open-source models (Llama, Mistral, CodeLlama, etc.) through Ollama's unified API
5. Model-Agnostic Design: Aligns with ADK's model-agnostic philosophy mentioned in the README

Current State

• ADK Go currently implements the model.LLM interface only for Gemini models
• The model/gemini package provides a reference implementation
• Models are created via gemini.NewModel() and passed to agents

[aviralgarg05]

I would like to solve this issue

[YootTanA]

Hey, this might be a silly question, but have you tried integrating Ollama with the adk via LiteLLM example. Don't get me wrong, I'd loved to have a native support from the framework. However, I'm wondering whether you've run into any issues - such as response latency, in case you have already tried it.

[byebyebruce]

checkout this https://github.com/byebyebruce/adk-go-openai