rig-core 0.44.0

An opinionated library for building LLM powered applications.
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
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87

//! Ollama configuration, model identifiers, and its chat, embedding and
//! model-listing wires.
//!
//! Chat goes through the daemon's OpenAI-compatible `/v1/chat/completions`
//! on the shared Chat Completions wire, as the
//! [`OLLAMA`](crate::providers::openai::wire::OLLAMA) dialect.
//! [`Ollama::native_completion`] reaches its native `/api/chat` instead
//! ([`Chat`]), the only route that sends model `options` such as `num_ctx`.
//! Embeddings use `/api/embed`, and the model listing `/api/tags`.
//!
//! ```no_run
//! use rig_core::providers::ollama;
//!
//! # fn run() -> Result<(), Box<dyn std::error::Error>> {
//! let provider = ollama::Ollama::new();
//!
//! let qwen = provider.completion("qwen2.5:14b");
//! let native = provider.native_completion("qwen2.5:14b");
//! let embeddings = provider.embedding(ollama::ALL_MINILM, Some(384));
//! # Ok(())
//! # }
//! ```
//!
//! `Ollama::from_env` reads `OLLAMA_API_BASE_URL` and `OLLAMA_API_KEY` for
//! remote or authenticated daemons.
pub mod chat;
pub mod embedding;
pub mod extension;
pub mod model_listing;
pub mod streaming;
pub mod wire;

pub use crate::client::ollama::Ollama;
pub use chat::Chat;
pub use embedding::{
    ALL_MINILM, BGE_M3, EMBEDDINGGEMMA, EmbeddingResponse, Embeddings, MXBAI_EMBED_LARGE,
    NOMIC_EMBED_TEXT, QWEN3_EMBEDDING,
};
pub use model_listing::{ListModelEntry, ListModelsResponse, Models};
pub use wire::OllamaConfig;

/// The address of a local daemon.
const OLLAMA_API_BASE_URL: &str = "http://localhost:11434";

/// Stable descriptor name recorded on normalized responses, streams, and
/// telemetry spans for this provider.
pub const PROVIDER_NAME: &str = "ollama";

/// The `llama3.2` model.
pub const LLAMA3_2: &str = "llama3.2";
/// The `llama3.1` model.
pub const LLAMA3_1: &str = "llama3.1";
/// The `llama3.3` model.
pub const LLAMA3_3: &str = "llama3.3";
/// The `llama4` multimodal model.
pub const LLAMA4: &str = "llama4";
/// The `llava` vision model.
pub const LLAVA: &str = "llava";
/// The `mistral` model.
pub const MISTRAL: &str = "mistral";
/// The `mistral-small3.2` model.
pub const MISTRAL_SMALL3_2: &str = "mistral-small3.2";
/// The `gemma3` model.
pub const GEMMA3: &str = "gemma3";
/// The `gemma4` model.
pub const GEMMA4: &str = "gemma4";
/// The `qwen3` model.
pub const QWEN3: &str = "qwen3";
/// The `qwen3.5` model.
pub const QWEN3_5: &str = "qwen3.5";
/// The `qwen3.6` model.
pub const QWEN3_6: &str = "qwen3.6";
/// The `qwen3.8` model.
pub const QWEN3_8: &str = "qwen3.8";
/// The `qwen3-coder` model.
pub const QWEN3_CODER: &str = "qwen3-coder";
/// The `deepseek-r1` reasoning model.
pub const DEEPSEEK_R1: &str = "deepseek-r1";
/// The `deepseek-v3.1` model.
pub const DEEPSEEK_V3_1: &str = "deepseek-v3.1";
/// The `gpt-oss` model.
pub const GPT_OSS: &str = "gpt-oss";
/// The `phi4` model.
pub const PHI4: &str = "phi4";

#[cfg(test)]
mod history_tests;