LatentKit is a route-based AI gateway giving one API for chat, images, audio, and video across 11+ AI providers with fallback, budgets, and tracing.
LatentKit is an AI control plane and API gateway that lets developers route requests across 11+ connected AI providers (plus OpenAI-compatible endpoints) through one unified base URL and authorization model. Teams can switch providers or models at runtime via routing policies without redeploying application code, and LatentKit automatically fails over to another provider if one has an outage.
The platform includes budget controls (monthly spend caps per API key), live request tracing with latency and cost metrics, and guardrails covering prompt controls, PII handling, and anomaly detection. Developers can use official SDKs for JavaScript, Python, and PHP, a CLI for terminal testing and CI integration, and pre-built integrations for WordPress, Drupal, Laravel, and the Vercel AI SDK.
LatentKit supports both bring-your-own-key (BYOK) usage and managed "Platform Access" where LatentKit brokers provider access directly. Larger teams can use white-label options such as custom domains and branded dashboards, along with audit logs and governance controls for managing multiple client workspaces.
LatentKit is a route-based AI gateway that gives applications a single API for chat, completions, embeddings, images, vision, audio, and video across 11+ connected AI providers.
Yes. The Free plan includes 1 app, 1 policy, unlimited BYOK (bring your own key) usage under fair use, and one Platform Access provider.
Yes. LatentKit routing policies let you assign or change which provider/model handles requests at runtime, without a code redeploy.
LatentKit automatically fails over to another connected provider to keep requests flowing.