These three providers all work with LuciQA, but they are not interchangeable accounts. Each needs its own API key. The most useful comparison is which models your account can access, how images are handled, and what quota applies today.
Gemini
Gemini uses Google’s API directly. LuciQA lists several Flash and Flash-Lite models, and its image path sends the selected PNG as image input. Google’s free tier is tied to project and model, so the active numbers belong in AI Studio rather than a permanent blog table. Check the provider’s data-use terms for your tier before sending private content.
Groq
Groq offers hosted inference with model-specific free-plan limits. LuciQA uses GPT-OSS for text by default and a Qwen vision model for image questions. If you select text and then send a diagram, the app changes the request model for that question. Inspect the actual model and limit shown in your Groq account.
OpenRouter
OpenRouter exposes multiple model endpoints behind one account. LuciQA’s current model choices are two free Gemma variants, both configured for text and images. The :free suffix matters. Free endpoints share account limits and may have limited capacity, so a successful key creation does not guarantee an uninterrupted session.
Which to choose?
For clear text, first try a provider whose current account limit you understand. For diagrams, test image behavior on a practice question. If you have Pro, use two providers when an independent comparison helps; two providers can still run related model families, so agreement is not proof. LuciQA’s self-reported confidence is not a verified accuracy score.
See the provider setup guides and free API comparison. Provider facts were last checked on 24 September 2026.