Bring Your Own Key (BYOK)

Bring Your Own Key (BYOK) is an Enterprise-only capability that lets your organization plug a Google Gemini API key directly into i1n. AI translations then run against your own Google AI account instead of the i1n shared inference pool.

When BYOK is active, your Enterprise contract receives a +50% bonus on AI translation credits and translation keys, on top of whatever limits were negotiated. Inference costs charged by Google are billed to your own Google account; i1n continues to handle orchestration, caching, variable protection, and brand voice enforcement.

BYOK is configured per organization from the dashboard. The key is encrypted at rest and is only used to make translation calls on your behalf — it is never logged, exported, or shared across organizations.

BYOK is worth enabling when any of the following apply: your organization already has a Google Cloud or Google AI Studio agreement, you require direct cost visibility on the underlying inference, or you want first-class access to the latest Gemini preview models as soon as Google releases them.

The +50% bonus on credits and keys means a BYOK-enabled Enterprise tenant effectively pays for orchestration only, while inference costs flow through your existing Google billing relationship — typically resulting in a lower total cost of ownership at high translation volumes.

BYOK also gives you tighter governance. You can rotate the key, set quotas in Google AI Studio, and audit usage from a single Google billing dashboard alongside the rest of your AI workloads.

Once BYOK is enabled, you can choose the default Gemini model used for AI translation. The model can be changed at any time from the dashboard and applies to subsequent translation jobs.

Gemini 2.5 Flash-Lite is the recommended default. It offers the strongest balance between translation quality, latency, and cost for the majority of localization workloads.

Gemini 3 Flash (preview) is suited for higher-quality general translation when latency is still a priority. Gemini 3.1 Pro (preview) provides the highest available quality and is recommended for nuanced or high-stakes content. Gemini 3.1 Flash-Lite (preview) is the fastest option and is ideal for very large batches where throughput matters most.

Preview models are subject to change by Google. i1n tracks new Gemini releases and exposes them in the dashboard as they become available — no contract amendment is required to switch models.

Step 1 — Generate a Gemini API key. Sign in to Google AI Studio (aistudio.google.com), open API keys, and create a new key scoped to the project you want to use for billing.

Step 2 — Add the key to i1n. From the dashboard, open your organization settings, navigate to the BYOK section, paste the API key, and save. The key is encrypted at rest immediately upon submission.

Step 3 — Choose your default model. Pick one of the available Gemini models (see Available Models above). The selection takes effect for all subsequent translation jobs across the CLI, MCP server, and dashboard.

Step 4 — Verify. Run i1n push --translate against a small set of keys and confirm the job completes. The dashboard usage view shows whether each translation was processed via BYOK or the default i1n inference path.

To rotate the key, replace it in the dashboard at any time — in-flight jobs continue using the previous key until they complete. To disable BYOK entirely, remove the key from the dashboard; future translations revert to the default i1n inference path automatically.