User Guides
These guides cover common patterns and best practices for working with SandBase in production.
Available Guides
| Guide | Description |
|---|---|
| Model Routing | How SandBase routes requests across providers — fallback chains, priority-based routing, and capability-aware selection |
| Streaming | SSE protocol details, OpenAI vs Anthropic streaming formats, and client-side handling |
| Error Handling | Retry strategies, exponential backoff, circuit breaker patterns, and idempotency |
| Rate Limiting | Understanding rate limit headers, handling 429 responses, and high-throughput best practices |
| Billing & Pricing | Credit system, top-up flow, per-model pricing, usage tracking, and cost optimization |
Quick Tips
- Use streaming for chat interfaces — it reduces perceived latency significantly
- Implement retries with backoff — transient errors from upstream providers are normal
- Monitor your usage — the dashboard shows real-time token consumption and costs
- Choose models wisely — smaller models (GPT-4o Mini, Claude 3.5 Haiku) are 10-50x cheaper for simple tasks