Squadbase LLM API
A built-in data source that lets your Squadbase apps use LLMs without providing your own API key
Squadbase LLM API is a built-in data source that lets your apps call LLMs without you having to provide an API key. It appears in the AI category under Data sources > Add data source in the sidebar. Usage is billed as AI credits. (→ AI credits)
You can use it to build features such as:
- AI chatbots
- AI-generated summary reports
- AI-powered anomaly detection
Models
Each LLM API connection is set to one of the following four model tiers. Squadbase decides and updates which model actually powers each tier. Even when the model is swapped, you don't need to change your app's code.
| Tier | Use |
|---|---|
| Quick | Simple tasks and workloads that need high throughput |
| Normal | General-purpose tasks |
| Smart | Complex reasoning and workloads that need high-quality output |
| Embedding | Generating text embeddings for vector search |
Make it available in a project
You create an LLM API connection from the project's chat.
When AI starts on an implementation that needs an LLM, an Add an LLM API data source card appears in the chat. It also appears if you tell the chat you want to use the LLM API. The card is prefilled with a Name and Model suited to the use case. Change them if needed, then click Create. The connection is created and becomes available in that project.
You can switch a text generation model (Quick / Normal / Smart) after the data source is added. The change applies to every app that uses the connection. An embedding model can't be changed after the data source is added, so choose it when you create the connection.
Use your own API key
If you want to connect directly to a specific provider, you can connect with your own API key. The following providers are supported:
- Anthropic
- Gemini
- OpenAI
Choose the provider from the AI category under Data sources > Add data source in the sidebar, and enter your API key.