Support more models

[ericdallo]

To upvote this issue, give it a thumbs up. See this list for the most upvoted issues.

Currently supported providers:

WIP:

• Gemini: Support google LLM provider #50

What other models should ECA support?

[marlonjsilva]

I think it would be really good to add support for at least the following providers:

• Google Gemini (already in progress);
• DeepSeek;

In this way, we cover the biggest players in the market.

We could also later look to add support for cloud providers like Google Vertex, AWS Bedrock, Azure, etc.

[ericdallo]

Awesome @marlonjsilva!

About deepseek, I tessted ollama with deepseek-r1, even so do you think would be nice to integrate with deepseek API itself?

[marlonjsilva]

If you used the command ollama run deepseek-r1, you are running a Qwen3 8b fine-tuned model, and not their full-fledged 671b parameter model. Running a model at this size locally would require very specialized hardware.

[ericdallo]

makes sense, so we should support deepseek integration indeed

[ericdallo]

Now ECA support custom providers as well via customProviders setting, allowing to reach providers that have API schema similar to existing providers (like litellm hosts).
Check https://github.com/editor-code-assistant/eca/blob/master/docs/configuration.md#custom-llm-providers

Support famous models is still a nice to have, so leaving this one open.

[SqrtMinusOne]

I think supporting OpenRouter would be very helpful. They provide a lot of models.

[ericdallo]

@SqrtMinusOne yes, @zikajk is working on supporting openai-legacy api which will unblock this!