Does Lava work with Bedrock, Azure, or custom providers?

Custom Provider Support

Lava's gateway supports 38+ AI providers out of the box. But we also support custom configurations for teams with specific provider requirements.

AWS Bedrock

If you're using AWS Bedrock (for example, running Anthropic models through Bedrock with your AWS credits), Lava can work with your setup. There are two integration approaches:

Gateway mode — Lava routes calls through the gateway with your Bedrock credentials. We're actively expanding support for AWS authentication methods including OIDC.
POST request mode — You make Bedrock calls directly using your existing code and authentication, then send usage data to Lava via POST requests. This is the simplest path if you have a complex auth setup (like OIDC from your infrastructure).

Azure OpenAI

Similar to Bedrock, if you're using Azure's OpenAI service, you can use either gateway mode with your Azure credentials or POST request mode to report usage after the fact.

Custom or Self-Hosted Models

If you're running custom or self-hosted models (e.g., fine-tuned models, locally hosted LLMs), you can still use Lava for billing and credit management:

Upload your custom model definitions to Lava
Use POST requests to report usage from your custom models
Lava handles all the billing, wallet management, and analytics

SDK Compatibility

Lava works with popular AI SDKs including:

OpenAI SDK
Anthropic SDK
Vercel AI SDK
LangChain
And most other SDKs that make standard API calls

For specialized SDKs (like the Claude Agent SDK), we work with customers to find the best integration path. In many cases, the POST request approach is the simplest way to get started while we build deeper SDK-specific integrations.

Requesting Support for New Providers

If you need a provider that isn't currently supported, reach out to us. Our proxy architecture is designed for extensibility, and we add new providers regularly based on customer needs.