Google Gemini models in Big-AGI

Running Google Gemini in Big-AGI

Add your Google AI Studio key and use Gemini models at Google's own API rates. Big-AGI isn't a reseller: no markup and no intermediary, so Google bills you directly.

• Your key, your billing. Usage is billed by Google directly, nothing routes through us.
• Thinking control, matched to the generation. Newer Gemini models get discrete thinking levels, older ones a numeric token budget; Big-AGI always sends the right shape.
• Long context, multimodal, real image generation. Feed Gemini a full codebase, a video transcript, or a stack of research papers in one context, with images in the same chat, and generate images with real aspect-ratio and resolution controls under the names Google itself markets.
• Two tracks, pick per task. Gemini ships Pro for depth and long context, Flash for speed and volume. Both appear in the table below, so you choose per task, not per app.

Why not just the Gemini app?

The Gemini app is Google's model in Google's box. Big-AGI runs Gemini next to GPT, Claude, Grok, and anything else you've connected, side by side in one conversation, which the Gemini app isn't built to do.

It also surfaces controls the consumer app keeps hidden: Gemini's API lets you set the thinking level and temperature per request, real parameters Google ships, not settings we made up. Your keys stay in your browser instead of your Google account.

Billing shifts from monthly to per-token too. A light user can spend less in a month than a subscription costs, and a heavy user never runs into a usage cap, only a bill for tokens actually used. Beam parallel runs do use more tokens than a single reply, that part isn't free.

Your keys and your data

Turn on Direct Connection under Google's Advanced settings and the browser calls Google directly, bypassing the Big-AGI server, whenever your key is client-side and Google's API allows it. Your keys stay in your browser. Chats are stored locally first and sync only if you turn it on. The AI Inspector shows the exact request before it goes out: model, parameters, token counts, and a cost estimate.

Gemini in Beam

Beam runs Gemini against GPT, Claude, or any other model you've connected, in parallel, then hands you Fusions: several strategies that combine, cross-check, and synthesize the parallel answers, which beats just picking the best one. Parallel runs use more tokens than a single chat.