aimux-core 0.1.0

Core abstractions for aimux: LanguageModel trait, message types, and user-facing API
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42

//! The `LanguageModel` trait — the provider-facing interface.
//!
//! Every LLM provider implements this trait. Users never call `do_generate` /
//! `do_stream` directly; they call `generate_text` / `stream_text` instead.

use async_trait::async_trait;

use crate::error::AiMuxError;
use crate::options::CallOptions;
use crate::result::{GenerateResult, StreamResult};

/// The unified language model trait (provider-facing).
///
/// Aligned with Vercel AI SDK `LanguageModelV4`.
///
/// # Implementation notes
///
/// - `do_generate` / `do_stream` are **not** user-facing API. Users call
///   `generate_text` / `stream_text` (free functions in [`crate::generate`]).
/// - Providers must emit `StreamStart` as the first stream part, and `Finish`
///   as the last (before any `Error`).
/// - Unsupported `CallOptions` fields must produce a `Warning::Unsupported`
///   rather than being silently dropped.
#[async_trait]
pub trait LanguageModel: Send + Sync {
    /// Specification version (always `"v4"`).
    fn specification_version(&self) -> &'static str {
        "v4"
    }

    /// Provider name, e.g. `"openai"`.
    fn provider(&self) -> &str;

    /// Model identifier, e.g. `"gpt-4o"`.
    fn model_id(&self) -> &str;

    /// Generate a complete (non-streaming) response.
    async fn do_generate(&self, options: &CallOptions) -> Result<GenerateResult, AiMuxError>;

    /// Generate a streaming response.
    async fn do_stream(&self, options: &CallOptions) -> Result<StreamResult, AiMuxError>;
}