Model Capabilities
Every model the platform runs, live — sources, fan-out queries, and per-market availability
The cards below render live from the public GET /llm/models endpoint — the exact catalogue the product exposes. What you see here is what your analyses run on.
- Sources & citations — the model returns the URLs behind its answer; these feed every citation metric.
- Fan-out queries — the intermediate search queries the model generated while answering. Only some models expose them.
- Market chips: US verified end-to-end · EEA answers arrive without sources · IT hard-blocked, requests do not execute.
- No chips = available in every market.
Interface models
Proxied or scraped consumer surfaces (ChatGPT, Copilot, Google AI Overviews…) rather than provider APIs — regional availability matters most here.
Native API models
Direct provider APIs — consistent worldwide, differences are in what they return.
Known regional limitations
- Gemini Interface no sources — EEA markets (including
FR) answer without sources; onlyUSis verified with full citations. - Google AI Mode — available in
FRwith full sources since July 22, 2026 (previously hard-blocked); 17 markets verified includingFR/DE/IT/ES/NL. - Google AI Overviews rolling out — live in
FRsince July 22, 2026, but Google is expanding coverage gradually over the first days, so French results may be sparse initially. - Copilot Interface hard block —
ITandBRare blocked; 15 other markets return sources.
The catalogue above is the full platform surface; which models run on your analyses depends on your subscription. Manage the selection per domain under Settings → AI Models in the app.
Exact response schema: GET /llm/models API reference.
Running shopping analyses? The AI shopping surfaces — ChatGPT and Alexa Shopping — are documented separately on Shopping Models.