chat-responses 0.1.2

Generic OpenAI Responses API client for chat-rs. Bring your own server.
Documentation

chat-responses

Generic OpenAI Responses API wire client for chat-rs. Symmetric counterpart to chat-completions: owns the /responses endpoint wire types and SSE stream parsing. Point it at any server implementing the Responses API.

Install

[dependencies]
chat-core = "0.3.0"
chat-responses = "0.1.2"
tokio = { version = "1", features = ["macros", "rt-multi-thread"] }

Or via the umbrella crate: chat-rs = { version = "0.4.0", features = ["responses"] }.

Usage

use chat_responses::ResponsesBuilder;
use chat_core::{builder::ChatBuilder, types::messages};

let client = ResponsesBuilder::new()
    .with_base_url("https://api.openai.com/v1")
    .with_model("gpt-4o")
    .with_api_key("sk-...")
    .build();

let mut chat = ChatBuilder::new().with_model(client).build();

let mut msgs = messages::from_user(vec!["Hello!"]);
let response = chat.complete(&mut msgs).await?;

Capabilities

  • Completions — text generation with tool calling and structured output
  • Streaming — SSE token streaming (requires stream feature)
  • Reasoning effort — .with_reasoning_effort("medium") for reasoning models
  • Previous-response-id round-tripping — server-side context resume (default on)
  • Custom tool declarations — .with_tool_declaration(value) for wrappers that want to inject provider-specific tools as raw JSON

When to Use This vs chat-openai

Use chat-responses directly when you control the endpoint or are wrapping a Responses-API provider that isn't OpenAI. Use chat-openai when you want the OpenAI defaults (URL + OPENAI_API_KEY), the OpenAI-specific native tools (web_search, image_generation), and the /embeddings endpoint.

Decoupled Native Tools

The wire layer is trait-agnostic. Wrappers materialize their provider-specific native tools into serde_json::Value and pass them via .with_tool_declaration():

let client = ResponsesBuilder::new()
    .with_base_url("https://api.openai.com/v1")
    .with_model("gpt-4o")
    .with_api_key("sk-...")
    .with_tool_declaration(serde_json::json!({ "type": "web_search" }))
    .build();

Custom Transport

Pluggable via .with_transport(). Defaults to HTTP/SSE via reqwest. WebSocket transports (AsyncWsTransport, WsTransport) work when paired with a Responses-API WS endpoint.

Feature Flags

Streaming is gated on the stream feature:

chat-responses = { version = "0.1.2", features = ["stream"] }