hermes - 💡(How to fix) Fix [Feature]: Support for Native Google / Vertex AI Provider (Bypass OpenRouter 402 Errors & Rate Limits) [3 comments, 3 participants]

I use the Hermes agent extensively with the `google/gemini-3.1-pro-preview` model, currently routed through the `openrouter` provider. 
However, I frequently encounter `HTTP 402` and API Rate Limits from OpenRouter (attempting to charge a markup) because Hermes tries to allocate large context windows (like `max_tokens: 65536`), but the OpenRouter credit balance check blocks the request natively, despite the underlying Google model being inexpensive or when I have sufficient Google Cloud credits available.


API call failed (attempt 1/3): APIStatusError [HTTP 402]
Provider: openrouter Model: google/gemini-3.1-pro-preview
Error: HTTP 402: This request requires more credits, or fewer max_tokens. You requested up to 65536 tokens, but can only afford 3312.


I maintain dedicated Google Cloud billing and have my own authentication (`google-creds.json` and API Keys), so routing my requests through OpenRouter introduces an unnecessary middleman, markup fees, and artificial rate limit bottlenecks.

Allow configuration of a native, direct `google` or `vertex-ai` provider in `config.yaml`, completely bypassing OpenRouter. 

1. **Google AI Studio / API Key Integration:** Use standard Google API keys as you do for `anthropic` or `openai`.
2. **Vertex AI Credentials Integration:** Allow defining a path to the Google Credentials JSON (`google-creds.json`) for corporate/project-based Vertex AI authentication. 

I'd like to configure my `config.yaml` to point my model `google/gemini-3.1-pro-preview` directly to Google's endpoints to fully utilize my Google Cloud credit allowance without OpenRouter intercepting the usage/limits.