chat-completions

Generic OpenAI Chat Completions wire client for chat-rs. Targets the /v1/chat/completions wire spec — point it at any OAI-compatible server (vLLM, llama.cpp, LiteLLM, Together, Fireworks, your own gateway).

Install

[dependencies]
chat-core = "0.4.0"
chat-completions = "0.2.4"
tokio = { version = "1", features = ["macros", "rt-multi-thread"] }

Or via the umbrella crate: chat-rs = { version = "0.5.0", features = ["completions"] }.

Usage

use chat_completions::ChatCompletionsBuilder;
use chat_core::{builder::ChatBuilder, types::messages};

let client = ChatCompletionsBuilder::new()
    .with_base_url("http://localhost:8000/v1")
    .with_model("my-model")
    .with_api_key("sk-...")  // optional — omit for servers that don't require auth
    .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?;

Bring your own base URL, model name, and (optional) API key.

Capabilities

• Completions — text generation with tool calling and structured output
• Streaming — token-by-token output (requires stream feature)
• Embeddings — vector embeddings (where the server supports /embeddings)

When to Use This vs a Dedicated Wrapper

Use chat-completions directly when you control the endpoint or want maximum flexibility. Use a dedicated wrapper (chat-ollama, chat-huggingface, chat-cerebras, chat-deepseek) when you want preset URLs, env var conventions, and provider-specific niceties.

Custom Transport

Supply a custom transport via .with_transport() to use something other than the default HTTP (reqwest):

let client = ChatCompletionsBuilder::new()
    .with_base_url("http://localhost:8000/v1")
    .with_model("my-model")
    .with_transport(my_transport)
    .build();

Feature Flags

Streaming is gated on the stream feature:

chat-completions = { version = "0.2.3", features = ["stream"] }