The simplest API to get structured data from any LLM
l1m is the easiest way to get structured data from unstructured text or images using LLMs. No prompt engineering, no chat history, just a simple API to extract structured json from text or images.
l1m provides a schema-first API that accepts JSON Schema definitions and returns structured data matching that schema. Users define their desired output structure once and the API handles prompt engineering automatically, eliminating the need for manual prompt crafting or multiple API calls.
l1m includes built-in caching with configurable TTL headers (x-cache-ttl). Identical requests with the same input, schema, and model are served from cache, reducing redundant LLM provider calls and lowering overall API costs.
l1m supports any OpenAI-compatible provider and Anthropic models through flexible headers (x-provider-url, x-provider-key, x-provider-model). Users can switch providers or use local models like Ollama by changing headers, maintaining the same API interface.
l1m accepts both base64-encoded images (JPEG, PNG) and text as input through the same /structured endpoint. Users can process both modalities with identical schema definitions and API calls.
l1m supports enum properties in JSON Schema to route requests to different tools or functions. Users can make an initial call to determine the appropriate tool, then extract tool-specific arguments in follow-up calls using the same API.