Guides
These guides cover common patterns for using SandBase in production.
Call models
| Guide | Description |
|---|---|
| Chat Completions | Call SandBase through the OpenAI-compatible chat protocol |
| OpenAI Responses | Call GPT models through the OpenAI-compatible Responses protocol |
| Anthropic Messages | Call SandBase through the Anthropic-compatible messages protocol |
| Google Gemini | Call SandBase through the Google Gemini-compatible GenerateContent protocol |
| Image and video | Generate images and videos through the shared asynchronous Run API |
| Streaming | Stream compatible Model output to your application |
| API Catalog | Browse API capabilities and their vendor-qualified endpoints |
Run in production
| Guide | Description |
|---|---|
| Errors and retries | Classify failures and retry only when an operation is safe to repeat |
| Rate limits | Handle 429 responses and smooth request bursts |
| Concurrency limits | How many tasks your organization can run at once |
| Pricing and usage | Understand credits, usage, and cost controls |
Quick Tips
- Use Setup when the user wants tools inside Codex, Claude, Cursor, or Kiro.
- Use Models or APIs when your app needs one capability.
- Use Build Agent when the work has multiple steps or should be reused.
- Use Activities in the Console to review requests and usage together.