litellm-rs

A high-performance Rust library and gateway for calling LLM APIs in an OpenAI-compatible format. Ships with 50+ built-in OpenAI-compatible providers plus first-class adapters for OpenAI, Anthropic, AWS Bedrock, Mistral, and Cloudflare.

Features

• 60+ runtime-wired providers - OpenAI, Anthropic, AWS Bedrock, Mistral, Cloudflare, plus 50+ OpenAI-compatible providers via the Tier 1 catalog. See Provider Support for the full matrix.
• OpenAI-Compatible API - Drop-in replacement for OpenAI SDK
• High Performance - 10,000+ requests/second, <10ms routing overhead
• Intelligent Routing - Load balancing, failover, cost optimization
• Enterprise Ready - Auth, rate limiting, caching, observability

Quick Start (5 Minutes, API-Only Recommended)

Most users use this project as a unified API library, not as a gateway server. Start with API-only mode first.

[dependencies]
litellm-rs = { version = "0.5", default-features = false, features = ["lite"] }

For crate users, no make is required.

Usage

As a Library (API Integration)

use litellm_rs::{completion, user_message, system_message};

#[tokio::main]
async fn main() -> Result<(), Box<dyn std::error::Error>> {
    let response = completion(
        "gpt-4",
        vec![
            system_message("You are a helpful assistant."),
            user_message("Hello!"),
        ],
        None,
    ).await?;

    println!("{}", response.choices[0].message.content.as_ref().unwrap());
    Ok(())
}

As a Gateway Server

Run from source repository

git clone https://github.com/majiayu000/litellm-rs.git
cd litellm-rs
cp config/gateway.dev.yaml.example config/gateway.yaml
cargo run --bin gateway

Install binary and run

cargo install litellm-rs --bin gateway
mkdir -p config
curl -L https://raw.githubusercontent.com/majiayu000/litellm-rs/main/config/gateway.dev.yaml.example -o config/gateway.yaml
gateway

Notes:

• gateway requires the storage feature at build time.
• Default features include sqlite, so default cargo run/cargo install satisfy this requirement.
• The development config starts without provider credentials or auth secrets and uses the local vllm catalog provider. Use config/gateway.yaml.example for production-style deployments with real provider keys and auth enabled.

Router Configuration

The gateway router config maps these fields into the runtime router:

• router.strategy selects the deployment routing strategy.
• router.circuit_breaker.failure_threshold controls consecutive failures before cooldown.
• router.circuit_breaker.recovery_timeout controls cooldown duration in seconds.
• router.circuit_breaker.min_requests sets the sample size required before cooldown.
• router.circuit_breaker.success_threshold sets the successes required to recover from cooldown.
• router.load_balancer.health_check_enabled enables pre-call deployment health checks.

router.load_balancer.sticky_sessions and router.load_balancer.session_timeout are reserved for future session affinity. Non-default values fail config validation until runtime affinity is implemented.

Installation

# Full gateway with SQLite + Redis (default)
[dependencies]
litellm-rs = "0.5"

# API-only - lightweight, no actix-web/argon2/aes-gcm/clap
[dependencies]
litellm-rs = { version = "0.5", default-features = false }

# API-only with metrics
[dependencies]
litellm-rs = { version = "0.5", default-features = false, features = ["lite"] }

# Gateway modules in library context (not standalone gateway binary runtime)
[dependencies]
litellm-rs = { version = "0.5", default-features = false, features = ["gateway"] }

Provider Support

Providers are organised into two tiers (see CLAUDE.md → Provider Tiers for the engineering definition).

• Tier 1 — catalog-only: OpenAI-compatible endpoints declared as data in src/core/providers/registry/catalog.rs. Routed through OpenAILikeProvider. Always available (no cargo feature required). The current crate runtime exposes chat completions and chat streaming for these providers; embeddings, images, audio, and other non-chat endpoints are not forwarded yet.
• Tier 2 — code-based: providers with custom request/response handling, auth signing, or streaming. Wired into the Provider enum and the factory. Some Tier 2 builders are feature-gated.

The matrix below is hand-maintained and reflects the runtime surface today. The source of truth for Tier 1 entries is catalog.rs; Tier 2 wiring lives in src/core/providers/factory/registry.rs. Capability columns describe which endpoints this crate exposes for the provider — passthrough means an implemented crate endpoint forwards the call to the upstream OpenAI-compatible endpoint without per-provider transformation. A dynamically-generated matrix is tracked as a follow-up.

Tier 2 — code-based providers

Provider	Cargo feature	Chat	Stream	Embed	Image	Audio	Notes
OpenAI (openai)	always	✅	✅	✅	✅	✅	Reference implementation.
Anthropic (anthropic)	always	✅	✅	–	–	–	Native Anthropic messages API.
Mistral (mistral)	always	✅	✅	passthrough	–	–	Native client.
Cloudflare Workers AI (cloudflare)	always	✅	–	–	–	–	Native client with account-id auth; streaming and embeddings currently return NotSupported.
Azure OpenAI (azure)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extra, but the factory path uses OpenAILike chat/stream only.
Azure AI Inference (azure_ai)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extra, but the factory path uses OpenAILike chat/stream only.
AWS Bedrock (bedrock)	always	✅	✅	✅	helper API	–	Native AWS Bedrock runtime path with SigV4 signing. Use openai_compatible for Bedrock Access Gateway or other OpenAI-compatible proxies.
Google Vertex AI (vertex_ai)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extra, but the factory path uses OpenAILike chat/stream only.
Meta Llama API (meta_llama)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extra, but the factory path uses OpenAILike chat/stream only.
Vercel v0 (v0)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extra, but the factory path uses OpenAILike chat/stream only.
Amazon Nova (amazon_nova)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extended, but the factory path uses OpenAILike chat/stream only.
fal.ai (fal_ai)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extended, but the factory path uses OpenAILike chat/stream only.
Replicate (replicate)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extended, but the factory path uses OpenAILike chat/stream only.
GitHub Models (github)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extended, but the factory path uses OpenAILike chat/stream only.
GitHub Copilot (github_copilot)	wired via factory (OpenAILike)	✅	✅	–	–	–	Native module retained behind providers-extended, but the factory path uses OpenAILike chat/stream only.
Generic OpenAI-compatible (openai_compatible)	always	✅	✅	–	–	–	For self-hosted / unlisted chat-completions endpoints.

Tier 1 — catalog providers (OpenAI-compatible, always available)

All entries below route through OpenAILikeProvider. Chat and streaming work for any endpoint that follows OpenAI's /chat/completions SSE protocol. Embeddings, images, audio, and other non-chat endpoints are not exposed through this path today, even when the upstream provider offers them.

Cloud (Bearer auth via env var):

groq, together, fireworks, perplexity, cerebras, openrouter, deepinfra, deepseek, novita, nvidia_nim, nebius, nscale, hyperbolic, featherless, galadriel, sambanova, heroku, friendliai, xai, moonshot, dashscope, qwen, baichuan, minimax, volcengine, xiaomi_mimo, zhipu, lemonade, linkup, poe, wandb, nanogpt, aiml_api, aleph_alpha, anyscale, bytez, comet_api, compactifai, maritalk, siliconflow, yi, lambda_ai, ovhcloud

Local (no API key):

vllm, hosted_vllm, lm_studio, llamafile, docker_model_runner, xinference, infinity, oobabooga

Experimental / module-only

The following modules exist under src/core/providers/ (gated on providers-extra or providers-extended) but are not wired into the unified Provider enum or the factory today. They compile but cannot be selected through create_provider/from_config_async. Treat them as experimental scaffolding subject to change:

ai21, baseten, clarifai, codestral, cohere, custom_api, databricks, datarobot, deepgram, deepl, elevenlabs, empower, exa_ai, firecrawl, gemini, gigachat, google_pse, gradient_ai, huggingface, jina, langgraph, manus, milvus, morph, nlp_cloud, oci, ollama, petals, pg_vector, predibase, ragflow, recraft, runwayml, sagemaker, sap_ai, searxng, snowflake, spark, stability, tavily, topaz, triton, vercel_ai, voyage, watsonx

For self-hosted or unlisted OpenAI-compatible endpoints, prefer the generic openai_compatible provider type instead.

Environment Variables

# Provider API Keys
OPENAI_API_KEY=sk-...
ANTHROPIC_API_KEY=sk-ant-...
GOOGLE_API_KEY=...
AZURE_OPENAI_API_KEY=...
AWS_ACCESS_KEY_ID=...
AWS_SECRET_ACCESS_KEY=...
AWS_REGION=us-east-1
GROQ_API_KEY=...
DEEPSEEK_API_KEY=...
MOONSHOT_API_KEY=...
ZHIPU_API_KEY=...
MINIMAX_API_KEY=...

# Optional
LITELLM_VERBOSE=true  # Enable verbose logging

Examples

Multi-Provider Routing

use litellm_rs::{completion, user_message};

// Automatically routes to the right provider based on model name
let openai = completion("gpt-4", vec![user_message("Hi")], None).await?;
let anthropic = completion("anthropic/claude-3-opus", vec![user_message("Hi")], None).await?;
let groq = completion("groq/llama-3.1-8b-instant", vec![user_message("Hi")], None).await?;
let bedrock = completion(
    "bedrock/us.anthropic.claude-3-5-sonnet-20241022-v2:0",
    vec![user_message("Hi")],
    None,
)
.await?;

bedrock/ uses the native AWS Bedrock provider. It signs requests with AWS SigV4 and preserves AWS execution model IDs such as us.*, global.*, region-prefixed IDs, and Bedrock ARNs. Use openai_compatible for Bedrock Access Gateway or other OpenAI-compatible proxies instead.

Embeddings

use litellm_rs::{embedding, embed_text};

// Single text
let embedding = embed_text("text-embedding-3-small", "Hello world").await?;

// Batch
let embeddings = embedding(
    "text-embedding-3-small",
    vec!["Hello", "World"],
    None,
).await?;

Streaming

use litellm_rs::{completion_stream, user_message};
use futures::StreamExt;

let mut stream = completion_stream(
    "gpt-4",
    vec![user_message("Tell me a story")],
    None,
).await?;

while let Some(chunk) = stream.next().await {
    if let Ok(chunk) = chunk {
        print!("{}", chunk.choices[0].delta.content.unwrap_or_default());
    }
}

Performance

• Throughput: 10,000+ requests/second
• Latency: <10ms routing overhead
• Memory: ~50MB base footprint
• Concurrency: Fully async with Tokio

Troubleshooting

Build/test uses too much CPU or memory

• Use API-only defaults first: cargo test --lib --tests --no-default-features --features "lite"
• Limit local parallelism when needed: CARGO_BUILD_JOBS=4 cargo test --lib --tests --no-default-features --features "lite" -- --test-threads=4
• Avoid --all-features unless you are doing release/nightly validation

I only need provider API aggregation, not gateway

• Prefer default-features = false with features = ["lite"]
• Use gateway runtime commands only when you need HTTP server/auth/storage middleware

Documentation

• API Documentation
• Documentation Index
• Development Gateway Config
• Production Gateway Config
• Examples

Contributing

See CONTRIBUTING.md for development setup and guidelines.

Security

See SECURITY.md for security policy and vulnerability reporting.

License

MIT License - see LICENSE for details.

Acknowledgments

Inspired by LiteLLM (Python).