CapkiiEvery model, every protocol. One gateway.

OpenAI, Anthropic and Gemini APIs behind a single base URL. Point your tools at one address and use any model — self-hosted, with your own keys.

Get Started

Guide

Diagram of the gateway: members and apps on the left each hold a key and connect into the dashed organization boundary, which holds the Capkii gateway and a budget meter; two lanes on the right lead to cloud providers and organization-owned keys.

Multi-protocol compatibility

Exposes OpenAI-, Claude- and Gemini-compatible APIs at the same time. Existing applications switch to Capkii by changing only the base URL and the API key — no code changes required.

Organizations and multi-tenancy

Hierarchical management of organizations, members and roles. Members are billed independently against the organization quota, which fits teams, departments and multiple customers sharing a single deployment.

Budget and usage control

Budget caps at both the organization and member level, together with API key quota and expiry controls. Requests are blocked the moment a limit is exceeded, preventing unexpected spend.

Channel and provider management

Orchestrate channels across providers with group-based routing, model mapping and wildcards, retry on failure and automatic disabling, plus a per-channel proxy.

Flexible billing

Supports usage-based ratios and per-request pricing, extra ratios and bulk price maintenance. Billing rules are explicit and can be tuned per model and per scenario.

Observability

Request logs, call modalities, latency and spend details, combined with dashboards and usage analytics, make cost attribution and capacity planning straightforward.