Ailib
Ecosystem v0.4.0

AI-Protocol Ecosystem

One specification, multiple runtimes.

Unified AI model interaction across Protocol, Rust, and Python.

★ Protocol Spec Rust Runtime Python Runtime

Dual-licensed MIT/Apache-2.0 · Protocol-driven design · Vendor neutral

Core Values

Protocol-Driven

All logic is operators, all configuration is protocol. Decouple providers from code with declarative configuration.

Multi-Runtime

Choose Rust for performance or Python for flexibility. Both share the same unified specification.

Extensive Provider Support

30+ AI providers, including global and China-region services, accessible through a unified interface.

Production-Grade Reliability

Built-in retry, rate limiting, circuit breaker, and other enterprise-grade features.

Type-Safe

Rust compile-time checks + Python Pydantic runtime validation.

Observability

OpenTelemetry integration with comprehensive metrics and tracing.

Core Projects

AI-Protocol

v0.4.0

Specification for unified AI model interaction with declarative configuration and operator-based processing.

30+ ProvidersOperator-basedDeclarative Config
Learn more →

ai-lib-rust

v0.6.6

High-performance Rust implementation with 14 architectural layers, type safety, and <1ms overhead.

Type-safe<1ms overheadEnterprise-grade
Learn more →

ai-lib-python

v0.5.0

Official Python runtime with 95% feature parity, Pydantic v2 type safety, and async support.

PyPI publishedJupyter integration96% complete
Learn more →

Runtime Comparison

Choose the right runtime

Feature AI-Protocol Rust SDK Python SDK
Type System YAML/JSON Schema Compile-time check Runtime type check
Performance N/A <1ms overhead ~10-50ms
Ecosystem 30+ Providers Crates.io PyPI
Ideal For Protocol implementers Systems programming ML/Data Science

Supported Providers

28 AI providers, global coverage, unified interface

Global Providers 🌍

Leading international AI providers

OpenAI favicon OpenAI
Anthropic favicon Anthropic
Groq favicon Groq
Gemini favicon Gemini
Mistral favicon Mistral
Cohere favicon Cohere
OpenRouter favicon OpenRouter
Replicate favicon Replicate
TogetherAI favicon TogetherAI
Perplexity favicon Perplexity
AI21 favicon AI21
HuggingFace favicon HuggingFace
DeepSeek favicon DeepSeek
xAI Grok favicon xAI Grok
Azure OpenAI favicon Azure OpenAI

China Region 🇨🇳

Leading China-region AI providers

ZhipuAI favicon ZhipuAI
MiniMax favicon MiniMax
Qwen favicon Qwen
Moonshot Kimi favicon Moonshot Kimi
Baidu Wenxin favicon Baidu Wenxin
Tencent Hunyuan favicon Tencent Hunyuan
iFlytek Spark favicon iFlytek Spark
Baichuan favicon Baichuan
SenseNova favicon SenseNova

Local & Self-hosted 🏠

Self-hosted and custom models

Ollama favicon Ollama
vLLM favicon vLLM
LlamaCpp favicon LlamaCpp
LocalAI favicon LocalAI
View full provider list →

Performance & Methodology

The following metrics refer to SDK layer overhead only (excludes remote model latency). See repository README for full methodology. Always benchmark with your workload.

0.6–0.9ms Per-request overhead
<2ms Streaming parse cost
11K–13K Mock throughput req/s

Note: Real-world throughput constrained by provider rate limits & network. Figures are indicative, not guarantees.

Architecture Snapshot

Layered design: App → High-level API → Unified abstraction → Provider adapters → Transport (HTTP/stream + reliability) → Common types.

Architecture snapshot diagram: app, high-level API, unified abstraction, adapters, transport, common types
Schematic: click to enlarge.

FAQ

Does it support local inference? Yes via Ollama; can mix with cloud providers.
How to know if an error is retryable? Errors are typed; use e.is_retryable().
Do you log request content? No. Content is not logged by default; add metrics hooks if needed.
Is function calling supported? Yes unified through Tool + FunctionCallPolicy.
When should I NOT use it? One-off small scripts targeting a single provider—use the vendor SDK directly.
Build: 9d1a7b2 · 2026-02-07T17:06:12.662Z · v0.21