Your own API key: the complete guide

1. What an own key gives you

Generations that go through your key are not charged in tokens. The AI request log marks them "own key, not billed".

Your key belongs to your account, not to a project: it follows you into every project you work in.

In projects you own, teammates who have no key of their own run on yours (and on your tokens when you have some), so a team needs one key, not one per person.

Using an own key is unlocked by the "Own API key" subscription on the billing page, or by a platform administrator for your account. The subscription only unlocks the feature; Google bills the usage.

The same Gemini key covers text, images, Veo video, speech and Lyria music. Kling video and ElevenLabs audio use their own separate keys (section 4).

2. Where to get a Gemini API key

Open Google AI Studio at https://aistudio.google.com/ and sign in with a Google account.

Choose "Get API key" (the key icon in the left bar), then "Create API key". Pick an existing Google Cloud project or let AI Studio create one — the key lives in that project, and that project is what gets billed.

Copy the key right away and keep it private. It looks like "AIza…" and is shown in full only once; you can always create another one.

Recommended: in the Google Cloud console, restrict the key to the "Generative Language API" so it cannot be used for anything else if it leaks.

3. Enable billing — the step that actually matters

A new key is on Google's free tier. The free tier has ZERO quota for image and video models: every image generation is refused with "429 RESOURCE_EXHAUSTED … free_tier … limit: 0", however much you can do in the Gemini chat app or the AI Studio playground. Those are separate products with their own allowances; the API is billed to your Google Cloud project.

In AI Studio open Settings → Plan (or "Upgrade" next to the key) and switch the key's project to the paid tier by linking a Cloud Billing account with a card.

The key itself does not change. Quotas usually update within a few minutes.

Set a budget alert in Cloud Billing so a busy pipeline cannot surprise you.

AssetsLine checks this for you: when you save a key it sends one tiny text request to the image model — a few tokens, a fraction of a cent — and a free-tier key is rejected on the spot with an explanation instead of failing your first pipeline.

4. Adding the key in AssetsLine

Open Settings (the avatar menu at the top right) → "Your API keys", paste the Gemini key and press Save. The key is verified during the save.

The key is stored on our server for your runs only. It is never shown again in full and never revealed to other members; "Remove" deletes it immediately.

Kling video: a key from https://klingai.com/dev (an older Access Key / Secret Key pair also works). ElevenLabs sound effects and voices: a key from your ElevenLabs profile → API keys. OpenAI (GPT text models and GPT Image): a key from https://platform.openai.com/api-keys, with a funded account. OpenRouter (Claude, Grok and hundreds of other text and image models, GPT and Gemini included): a key from https://openrouter.ai/settings/keys, with credits on the account. Their models then appear in the model lists of the text and image nodes. The order in which tokens and keys are tried, and which of them are off, is set in Settings → API routing. Each provider you have no key for keeps running on tokens.

5. Tokens or key: the switch in the top bar

Next to your balance the top bar shows two segments: tokens (the bolt with your balance) and key. The highlighted one is what the next run will use.

Default: tokens first. While your balance — or, in a project you do not own, the owner's balance — has anything left, runs spend tokens. Once every balance is empty the run switches to a key automatically, so nothing stops.

Click the key segment to prefer the key. Runs then go through your key while it works; tokens become the fallback for a provider you have no key for.

Who pays in a shared project: the person who runs the node, first from their own tokens or key, then from the project owner's. The owner's balance and key cover any member.

6. Costs and limits on Google's side

Google bills per use at its published rates (https://ai.google.dev/pricing). As a guide: an image on the flash image models is a few US cents (roughly $0.04–0.07), a text prompt is a fraction of a cent, Veo video is priced per second and is by far the most expensive.

Rate limits: a paid project starts at Tier 1 with a fixed number of requests per minute and per day per model; the tiers rise with cumulative spend. A run that hits the per-minute limit gets a 429 with a retry hint — AssetsLine waits and retries up to six times before giving up.

Some models are not offered in every country (several image and preview models are unavailable in the EU, UK and Switzerland). A "model not found" or "not supported in your region" error on a model that works on our platform key means exactly that.

Google may refuse a prompt on safety grounds; such a call returns text instead of an image and is not charged by us.

Failed calls are never charged in tokens; on your own key Google's own rules apply.

7. Troubleshooting

"… is not available on your own Gemini API key: Google's free tier does not include it (limit 0)" — billing is not enabled for the key's project. Section 3.

"Gemini rejected this API key: API key not valid" — the key was copied incompletely or deleted in AI Studio. Create a new one.

Repeated 429 without "limit: 0" — you are at the per-minute or per-day limit of your tier. Wait, or run fewer nodes in parallel.

Generations still spend tokens — check the top-bar switch (section 5) and that the subscription is active on the billing page.

If you think a key has leaked, delete it in AI Studio first, then remove it in AssetsLine and create a fresh one.