aisuite

Overview

aisuite is Andrew Ng’s open-source Python library that solves a common developer pain point: every LLM provider ships its own SDK with its own calling conventions, response formats, and authentication patterns. aisuite wraps them all behind a single interface that follows the OpenAI chat completions convention. You specify the provider and model as a string like anthropic:claude-sonnet-4-20250514 or openai:gpt-4o, and the library routes the call to the right backend, returning a consistent response object regardless of which provider handled the request.

Key Capabilities

The core value proposition is simplicity. A developer can test the same prompt across Claude, GPT-4o, Gemini, and Mistral models by changing one string — no SDK swaps, no response format parsing, no authentication refactoring. The library supports tool calling and streaming where the underlying provider does, and its provider-agnostic design makes it straightforward to build fallback chains (try provider A, fall back to provider B on failure) or cost-optimized routing (send simple queries to cheaper models, complex ones to stronger models). The codebase is intentionally small and dependency-light, making it easy to audit and extend.

Use Cases

Developers use aisuite for multi-model evaluation — running the same test suite against several providers to compare quality, latency, and cost before committing to one. It is also useful for building resilient applications that can fail over between providers, and for rapid prototyping where you want to experiment with different models without rewriting integration code. Teams that need to support multiple LLM backends for different customers or regions find the unified interface particularly valuable.

Considerations

aisuite focuses on the chat completions interface and does not attempt to unify every provider-specific feature (embeddings, fine-tuning, batch APIs). Each provider’s API key must still be configured separately. The library adds a thin abstraction layer, so provider-specific optimizations or advanced features may require dropping down to the native SDK. As an open-source project, support comes from the community rather than a dedicated team.

Who It’s For

aisuite is ideal for developers and teams who work with multiple LLM providers and want a clean, consistent interface without vendor lock-in. It appeals to engineers doing model evaluation, building multi-provider applications, or simply prototyping quickly across different models. Anyone who has been frustrated by maintaining separate integration code for each LLM provider will appreciate the simplicity.