Mirascope

Mirascope is an open-source development library for building AI applications with local models or cloud providers. Its documented local backends include Ollama, vLLM and LM Studio, plus direct inference through MLX on Apple Silicon Macs. Developers can use a shared model interface while keeping inference on their own hardware or sending requests to OpenAI, Anthropic or Google.

Ollama has a built-in provider. Local servers such as vLLM and LM Studio connect through their OpenAI-compatible APIs. These local paths support private or offline model use; choosing a cloud provider sends model requests to that service instead.

The library supports streaming responses, asynchronous calls and multi-turn conversations. Tool calling lets an agent invoke application functions and continue with their results, with documented support for tools through Ollama and vLLM. This makes it useful for developers building agents that need access to their application's data or actions.

MLX loads models from Hugging Face and runs them on the Mac, downloading them on first use and retaining loaded models in memory for later calls. That backend supports text generation and streaming, but doesn't support tools or structured output.

Mirascope has Python and TypeScript implementations under the MIT license. Automatic versioning, tracing and cost tracking help developers inspect model calls, including messages and tool inputs and outputs.