Expand description
HTTP clients for streaming LLM responses with performance recording
This module provides HTTP clients that leverage async-openai with BYOT (Bring Your Own Types) feature to work with OpenAI-compatible APIs. The clients support recording streaming responses for performance analysis.
Structs§
- Base
Http Client - Base HTTP client with common functionality
- GenericBYOT
Client - Generic BYOT client using serde_json::Value for maximum flexibility
- Http
Client Config - Configuration for HTTP clients
- Http
Request Context - Context for HTTP client requests that supports cancellation This bridges AsyncEngineContext and reqwest cancellation
- Http
Response Stream - A wrapped HTTP response stream that combines a stream with its context This provides a unified interface for HTTP client responses
- NvCustom
Client - NV Custom client using NvCreateChatCompletionRequest with Annotated responses
- Pure
OpenAI Client - Pure OpenAI client using standard async-openai types
Enums§
- Http
Client Error - Error types for HTTP clients
Type Aliases§
- Byot
Http Response Stream - Type alias for HttpResponseStream with BYOT responses
- Byot
Response Stream - Type alias for generic BYOT response stream
- NvChat
Response Stream - Type alias for NV chat response stream
- NvHttp
Response Stream - Type alias for HttpResponseStream with NV chat completion responses
- OpenAI
Chat Response Stream - Type alias for pure OpenAI chat response stream
- OpenAI
Http Response Stream - Type alias for HttpResponseStream with pure OpenAI responses