Gemini API: native requests and a live catalog
Native Gemini calls use /v1beta/models/{model}:generateContent or streamGenerateContent. Only models with an enabled channel, a valid group and a public multiplier appear in the live catalog. Current model IDs, supported endpoints and group pricing are published in the live catalog.
Integration endpoint
http://localhost:3000/v1beta · /v1beta/models/{model}:generateContent
Routing boundary
Gemini has no dedicated smart group. When you create a key, choose a regular group that the live catalog lists for the model; do not use the smart group of another model family.
Models in this family (3)
Gemini models in the catalog snapshot of 2026-10-17 with their lowest tier price. Model pages are published in Chinese; the model catalog covers every family.
Gemini: 3 models; the lowest tier is 0.5x (offered for 3 models); that tier costs about 7.1% of the official CNY reference price (0.5 ÷ 7).
| Model ID | Lowest tier | Lowest-tier price |
|---|---|---|
| gemini-4-argon-high | 0.5x tier | Input CNY 1 · Output CNY 5 per 1M tokens |
| gemini-4-argon-low | 0.5x tier | Input CNY 1 · Output CNY 5 per 1M tokens |
| gemini-4-argon-medium | 0.5x tier | Input CNY 1 · Output CNY 5 per 1M tokens |
Gemini models in other families:
| Model ID | Lowest tier | Lowest-tier price |
|---|---|---|
| gemini-3-pro-image-preview | 0.2x tier | CNY 0.2 per request |
| gemini-3.1-flash-image-preview | 0.2x tier | CNY 0.2 per request |
Frequently asked questions
What is 懂api?
懂api is an AI API relay and model aggregation gateway for developers: one API key and one base URL give access to GPT, Claude, Gemini and other text models plus image and video generation models, with verifiable prices and per-request charges.
Where can I verify current models and prices?
Use the live catalog at /api/geo/catalog and keep its version and checked_at values with any current-model answer.
Last updated: