Our take
Google AI Studio is Google's first-party inference platform for the Gemini model family, offering 44 tracked variants with a wide range of throughput tiers that trade speed against cost. It holds a verified security attestation and does not train on prompts, though EU-region hosting and zero-retention remain unverified in our data.
Use this when you need maximum throughput on an image-capable model — one variant reaches 465 tps. Pick the entry tier of Gemini 2.5 Flash Lite for budget-conscious text generation, or any tier when SOC 2 and a no-training policy are required. Skip it if you need confirmed EU data residency, a verified zero-retention option, or consistently fast throughput across every model variant.
- Widest throughput range for a single model — Gemini 2.5 Flash Lite spans 31 tps up to 186 tps, a 6× spread with the fastest tier twice the price of the slowest.
- Highest measured throughput in the sample set: 465 tps on Nano Banana 2 Lite Gemini 3.1 Flash Lite Image, 3.8× the next fastest tier.
- Strong compliance posture: SOC 2 attestation held and explicit no-training policy verified.
- Broad first-party catalogue — 44 tracked offers across Gemini 2.5, 3, 3.1 and 3.5 families including Flash, Flash Lite and image variants.
- Some tiers are extremely slow: Gemini 2.5 Flash at 2 tps is 93× slower than the fastest measured tier and 37× slower than its own Flash Lite variant at the same price point.
- EU data residency and zero-retention option both unverified in our data.
- Image model throughput mostly unmeasured — only one image variant has verified throughput.
Point your tools here
https://generativelanguage.googleapis.com/v1beta/openai/Model IDs on this page are LLMap's own slugs, built for browsing and comparing — not what Google AI Studio's API expects. Its own model identifier lives in the docs above.
Privacy & data handling
These answers cover requests sent to Google AI Studio through OpenRouter, as OpenRouter records them, last read 59 min ago. For Google AI Studio's own API we hold no answer: check its terms. SOC 2 and the data processing agreement below are hand-checked, not part of that sync.
We have not dated this attestation or linked the report yet.
Prompt retention: 55 days.