I tried running perplexica by using gemini api : gemini-2.0-flash and openai free api but none of them running gives error of 429

[Titli007]

I tried running perplexica by using free api key of gemini and openai but none of them running

429 You exceeded your current quota, please check your plan and billing details. For more information on this error, read the docs: https://platform.openai.com/docs/guides/error-codes/api-errors.

does perplexica support any free version of api of any llm? if so please mention the name and the model name

[delfireinoso]

I've used Gemini with Good results, but I prefer local models

[NovaRouteAI]

The practical angle in "I tried running perplexica by using gemini api : gemini-2.0-flash and openai free api but" is usage attribution: the model bill only becomes actionable when it can be tied back to a user, feature, or workflow.

I would separate raw usage collection from cost presentation: first normalize tokens and provider metadata per request, then let dashboards group by user, workflow, or model tier.

This is close to what I am experimenting with: official Chinese models behind an OpenAI-compatible multi-model layer, with an emphasis on predictable usage and routing behavior. For Vane, do you need billing-grade accuracy here, or is operational usage visibility enough for the first version?