Gemini API: a key in two minutes

The fastest way to start working with an LLM for free: no card, no phone number, no waiting. All you need is a Google account.

Steps

1. Open aistudio.google.com and sign in with a Google account;
2. On the left — “Get API key” → “Create API key”;
3. Pick a project (or create a new one) — the key appears instantly;
4. Copy the key: it starts with AIza…;
5. Test it right in AI Studio (“Run” and the model settings are on the right).

Free tier limits (snapshot, Nov 2026)

• Flash models: ~10 requests a minute, ~1,500 a day;
• context up to 1M tokens — long documents go through whole;
• flagship models — billing only;
• limits are counted per project: several projects = separate limits.

The first request

Via curl — works with any OpenAI-compatible client too:

curl "https://generativelanguage.googleapis.com/v1beta/models/gemini-3.8-flash:generateContent?key=KEY" \
-H 'Content-Type: application/json' \
-d '{"contents":[{"parts":[{"text":"Hi!"}]}]}'

How to avoid hitting the limits

• cache responses for identical requests;
• for bulk jobs use Gemini CLI — it has its own daily cap of 1,000 requests;
• on a 429, retry with exponential backoff — the limit refreshes within a minute;
• if you need more — the $300 Google Cloud credit covers Vertex AI too.

The catch

Google may use free-tier prompts to improve its products. For sensitive data, enable billing or use Vertex AI with zero-retention.

Snapshot 05.11.2026 · the service card is in the catalog