# Cumulus Labs

> Cumulus Labs (Y Combinator W26) builds a unified inference platform for production AI. A single OpenAI-compatible API at https://api.cumuluslabs.io/v1 consolidates eight subsystems — gateway, per-workflow router, layered prompt/KV cache, request-level observability, continuous shadow evaluation, one-click LoRA fine-tuning, custom open-weight hosting, and the proprietary Ion inference engine on NVIDIA Grace and Blackwell GPUs (30-50% more throughput than stock vLLM/SGLang). Drop-in for the OpenAI, Anthropic, LangChain, LlamaIndex, and Vercel AI SDKs — change one line of config.

## APIs

- [Cumulus Inference Gateway (docs)](https://docs.cumuluslabs.io/): OpenAI-compatible inference gateway; base URL https://api.cumuluslabs.io/v1.
- [Getting Started](https://docs.cumuluslabs.io/getting-started/): install and authenticate.
- [Inference Overview](https://docs.cumuluslabs.io/inference/overview/): inference jobs and live servers.

## Docs

- [Documentation](https://docs.cumuluslabs.io/)
- [Inference Glossary](https://cumuluslabs.io/glossary)
- [About](https://cumuluslabs.io/about)
- [Blog](https://cumulus.blog/)

## Authentication

- Bearer API key: `Authorization: Bearer <CUMULUS_KEY>` against https://api.cumuluslabs.io/v1 (OpenAI-compatible).

## Artifacts

- [Authentication profile](authentication/cumulus-labs-authentication.yml)
- [API conventions](conventions/cumulus-labs-conventions.yml)
- [Conformance](conformance/cumulus-labs-conformance.yml)
- [Lifecycle](lifecycle/cumulus-labs-lifecycle.yml)

## Company

- [GitHub organization](https://github.com/cumulus-compute-labs)
- [LinkedIn](https://www.linkedin.com/company/cumuluscomputelabs/)
- [Y Combinator](https://www.ycombinator.com/companies/cumulus-labs)
- Terms: https://cumuluslabs.io/terms-of-service.pdf — Privacy: https://cumuluslabs.io/privacy-policy.pdf