Migrating to Spro takes minutes. Because Spro speaks the OpenAI API, you keep your code, your SDKs, and your workflows — you change where requests are sent and how requests are authenticated.
Migration guides
Migrate from OpenRouterSame OpenAI-compatible API — one URL change and remapped model IDsMigrate from Direct ProvidersMove from calling OpenAI, Anthropic, or Google SDKs directly
Coming from another OpenAI-compatible gateway? The same two-step pattern applies — point your client at https://spro.hridaai.com/v1 and use your Spro API key — then remap model IDs using the OpenRouter guide as a reference.
What stays the same
- The OpenAI SDK — same client, same methods, same streaming
- Your messages and prompts — no format changes
- Structured output, temperature, max_tokens — all parameters pass through
- Your agent frameworks — LangChain, LlamaIndex, OpenAI Agents SDK, and others keep working
What changes
Only two values:
| Before | After |
|---|---|
| Provider-specific base URL | https://spro.hridaai.com/v1 |
| Provider API key | Your Spro API key |
And one convention: models are addressed with provider prefixes — e.g. gpt-oss-20b, anthropic/claude-sonnet-4, z-ai/glm-5.3-flash. See Models.
What you gain
- One API for every provider — switch models with a string change
- Automatic routing and failover — no provider-specific retry code
- Security guardrails — PII redaction, prompt-injection detection, output validation
- Centralized observability and cost controls — requests, tokens, latency, and spend in one dashboard
