Bring Your Own API Keys

Managed mode vs BYO mode

By default, FatThumb runs in managed mode: each generation is billed against your monthly credit allowance (or lifetime credits). The credits cover the AI model cost plus infrastructure.

BYO mode connects your own OpenAI and/or Gemini API keys. When a key is active, generations using that model are routed through your key — the model cost goes directly to your API provider at their published rates, and no credits are deducted from your FatThumb balance.

BYO mode works on any plan — there is no plan gate on it. You only need at least one stored provider key and BYO mode switched on (below). The usual model access still applies, though: the paid gpt-image-2 model requires a paid plan even with your own key, while Gemini is available on the free plan.

Adding your keys

1. Navigate to Settings → Provider Keys in the sidebar.
2. Enter your OpenAI API key (starts with sk-...) to enable gpt-image-2 in BYO mode.
3. Enter your Gemini API key to enable gemini-3-pro in BYO mode.
4. Keys are encrypted at rest. FatThumb never exposes them in responses or logs.

You can add one or both keys. In BYO mode you can only generate with a model whose provider key you've stored — if you pick the OpenAI model with only a Gemini key saved, that generation is rejected. Add both keys to use both models.

Switching BYO mode on

Adding a key is not enough by itself. FatThumb has two modes: managed (the default — billed against your credits) and your own keys. In Settings → Provider Keys, switch the mode to use your own keys (this requires at least one stored key).

Once it's on, generations route through your key for the matching provider and no credits are deducted. If you later remove your last key, the mode automatically reverts to managed.

Cost estimate

OpenAI gpt-image-2 generates at roughly $0.04–$0.08 per image at standard quality and $0.08–$0.16 at high quality (as of mid-2026, subject to change — check the OpenAI pricing page for current rates). Gemini 3 Pro pricing follows Google's published image generation rates.

For creators generating 120+ thumbnails per month, BYO mode on the Lifetime plan typically costs a few dollars per month in API fees, vs. the equivalent managed credit cost.

Spend cap and safety

FatThumb enforces a server-side spend cap on BYO usage to prevent runaway cost if a key is misused. If a generation would exceed the cap, it is rejected with an error. The spend cap is applied per organization per billing period and resets monthly.

Security

Your API keys are stored encrypted in the Convex backend. They are never transmitted to the browser after initial entry — the editor never receives the raw key. FatThumb uses your key only to call the respective model API for thumbnail generation in your organization; no other uses.

If you suspect a key has been compromised, revoke it in your OpenAI or Google Cloud console immediately and remove it from FatThumb Settings → Provider Keys.