Together AI models in Big-AGI

Running Together AI in Big-AGI

Add your Together AI key and pick from hundreds of open models on Together's serverless infrastructure, priced per token with no subscription. Big-AGI adds no markup and no intermediary: the billing relationship runs directly between you and Together AI.

• Your key, your billing. Usage is billed by Together AI to your account.
• Fully dynamic catalog. Pricing, context length, and model labels all come straight from Together's live API, so new models, Llama, DeepSeek, Qwen, Kimi included, show up the moment Together ships them, no hand-maintained list required.
• Vision figured out for you. Together's API doesn't publish which models accept images, so Big-AGI infers vision support from known multimodal families and naming, and those models correctly show an image-upload option.
• Serious serving speed. Together runs its own optimized inference stack, so open models answer at closed-model latencies.

Why Big-AGI instead of the playground?

Together's playground is a quick way to try one model. Big-AGI turns your key into a full workspace: persistent chats, personas, and attachments, layered over the raw API. Run a Together model in Beam next to Claude, GPT, and Gemini, something the playground was never built to do, while the parameters, the key, and the chats stay under your control.

Your keys and your data

Turn on Direct Connection and the browser talks to Together AI directly, skipping the Big-AGI server, when your key is client-side and Together 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, the token counts, and a cost estimate.

Together AI in Beam

Fan a prompt out across a few of Together's open models, or set one against Claude, GPT, and Gemini. Fusions then combine, cross-check, and synthesize the parallel answers instead of just picking the best one. Parallel runs use more tokens than a single chat.