Where Portkey fits
Portkey is strongest as a control and observability layer over accounts you already run. If you have negotiated provider rates, need an open-source gateway you can self-host for data-residency or compliance reasons, or want first-class guardrails, prompt versioning and deep request tracing, Portkey is built for exactly that. It is an LLMOps platform first — the gateway is one part of a governance stack aimed at teams standardising how many services call many models.
Where Kunavo wins
Kunavo removes the provider accounts entirely. You top up one wallet and call Claude, Gemini, GPT, plus image, video and audio, at 30–70% under each provider's list price depending on the model — no Anthropic Console, no Google Cloud project, no per-provider invoice to reconcile. Generation endpoints (images, video, music) are first-class, the native Anthropic Messages API is available on the same key, and checkout is Stripe-native with local methods (Apple/Google Pay, ACH, SEPA, Alipay, WeChat Pay). For a team that wants the cheapest path to many models on one bill — rather than a governance platform over keys it already holds — Kunavo is the shorter route.
How to pick (or use both)
Pick Portkey if you need a self-hostable open-source gateway, deep guardrails and prompt governance over provider keys you already manage. Pick Kunavo if you want resold inference on one prepaid wallet at 30–70% under list, first-class multimodal, and one invoice. And they compose: add Kunavo as an OpenAI-compatible provider inside Portkey to keep Portkey's control plane while your tokens run through Kunavo's discounted catalog.