Our take
Google AI is the first-party host for Google's own Gemini models, offering a tightly curated catalogue with no third-party downloads. Its 14 tracked offers range from a cheap Flash Lite entry point to a high-capability Pro tier with steep output pricing.
Use this for native Gemini feature access such as grounding, multimodal inputs, or Google Cloud integration. Pick it for budget prototyping on Flash Lite variants, or for highest-capability Gemini tasks where Pro pricing is acceptable. Skip it if you need verified compliance attestations, a zero-retention option, or a broad choice of non-Google models.
- Cheapest entry point for first-party Gemini inference. Gemini 2.5 Flash Lite input is 12.5× cheaper than Gemini 2.5 Pro, and its output is 25× cheaper.
- Tight vertical integration: all 14 tracked offers are Google-native models, with Gemma 2 27B as the sole downloadable option.
- Clear tier separation between Flash and Pro families, with three distinct input and output price bands.
- Compliance and data governance are largely undocumented: SOC 2, zero-retention option, prompt-training policy, EU endpoint, and headquarters country are all unverified or not disclosed.
- No throughput or latency data to judge performance: unmeasured on all 14 offers.
- Narrow catalogue with steep Pro pricing and no third-party breadth beyond a single Gemma model.
Privacy & data handling
- ✓
- Yes
- ✗
- No
- No answer on record
The colour says whether the answer favours you, not whether it is a yes — not training on your prompts earns a green cross, no zero-retention option earns a red one.