The Dipole ML API Gateway: Native Dual Compatibility for OpenAI and Anthropic SDKs

The Dipole ML API gateway at api.dipoleml.com was built around a simple observation: developers should not have to rewrite working code to switch frontier models. The gateway speaks both the OpenAI and Anthropic SDK formats natively, so most teams migrate by changing one line.

Native dual compatibility

Point your existing OpenAI SDK at the chat completions endpoint, or your existing Anthropic SDK at the messages endpoint. Both are first-class citizens on the same gateway, serving the same Raven models:

• OpenAI format: /v1/chat/completions, drop-in for the official openai client libraries
• Anthropic format: /v1/messages, drop-in for the official anthropic client libraries

Streaming, tool execution, and multimodal vision work identically in both formats with sub-second first-token latency. There is no translation shim, no degraded compatibility mode: requests are handled natively on both paths.

Automatic prompt caching economics

Repetitive input is expensive, so the gateway caches it for you. When input hits our cache, prompt caching discounts apply automatically: cached input drops to $0.05 per 1M tokens on Raven Flash and $0.13 per 1M tokens on Raven Max, saving up to 90% on input costs. No code changes, no flags to set.

Prepaid credits, zero lock-in

The API is pay-as-you-go with prepaid credit packs at $5, $10, $20, and $50. Credits never expire, they are independent of any subscription, and each request is debited from actual token usage. Raven Flash bills $0.50 per 1M input and $2.50 per 1M output; Raven Max bills $1.25 per 1M input and $6.25 per 1M output.

Switch in minutes

Update your base URL, keep your client library, and start streaming. The developer overview summarizes the endpoints, and the full reference includes copy-ready quickstarts for both SDK families.