Portkey
Portkey is a gateway and control plane for language model traffic: one OpenAI-compatible endpoint across many models, with fallbacks, retries and caching.
Highlights
- A universal OpenAI-compatible API to more than 1,600 LLMs with unified key management
- Reliability routing: automatic fallbacks, load balancing, retries, and timeouts
- Observability: request logs, distributed traces, cost and latency metrics, and alerts
- A prompt-management studio with templating and versioning
- Guardrails: PII redaction and configurable input and output policy enforcement
- Governance: role-based access control, virtual keys, and per-team budgets
- Simple and semantic caching to cut cost and latency
- An MCP gateway for secure tool access, with an open-source core
External link — opens portkey.ai in a new tab. Portkey is a third-party product; we are not affiliated with it.
Link checked 17 November 2026: this site responded and still names the product.
About Portkey
What it is
Portkey is a gateway and operations control plane for language model traffic, giving one OpenAI-compatible endpoint across more than sixteen hundred models with fallbacks, load balancing, retries and caching, plus observability, guardrails and access governance over what passes through it.
Why it's different
It is among the more mature options in this category, and the reliability features are what distinguish it from a pure cost router. Automatic fallback when a provider degrades is the one that earns its place in production, because model providers do have bad hours and an application with a single hard dependency goes down with them. Caching is the quiet cost saving, since a surprising share of real traffic is repeated. The considerations are the ones every gateway carries and they are not small: every prompt and response passes through a third party, and the gateway becomes a single point of failure in front of providers that were already redundant. Self-hosting is available and is the answer for sensitive workloads.
How people use it
It is adopted by teams running AI in production who need reliability and governance rather than a way to experiment — typically after an outage or after a security review asks who can call what. The practical order is fallbacks and observability first, since those address the failures you have already had, then caching and governance. Check where request and response bodies are retained before routing anything sensitive.
Written by the n3os team. We are not affiliated with Portkey.
More Model Platforms & APIs tools
- AurikoFreemium
Auriko is an LLM gateway: one OpenAI-compatible endpoint in front of more than a dozen model providers, routing requests by price, latency and reliability.
- Actualyze AIPaid
Actualyze AI is a single OpenAI-compatible layer between an organisation's applications and the model providers that governs and secures every request.
- EtchedPaid
Etched builds Sohu, an application-specific integrated circuit designed to do one thing: transformer inference. It targets inference, not training.
- HirundoPaid
Hirundo removes the effect of particular training data, such as bad labels or copyrighted material, from an already-trained model without full retraining.
- Hydra HostPaid
Hydra Host rents dedicated, non-virtualised NVIDIA GPU servers for AI, HPC and data workloads, across a network of more than forty data centres.
- Inception LabsPaid
Inception Labs is a Stanford spinout building Mercury, a family of diffusion-based language models that refine the whole output in parallel.
Every Model Platforms & APIs tool in the directory →
Listed by how many tags they share with Portkey, then by name. Nobody pays to appear here.
This listing was written from public information, without Portkey’s involvement. If you own it and something here is wrong — or you would rather not be listed at all — email us and, once we have confirmed it is yours, we will correct or remove it.
Get the ones worth knowing about
We write one of these for every tool worth the trouble. Get the new ones, plus what we have found genuinely useful lately.
Your address goes to Buttondown, who send the emails on our behalf. One click unsubscribes, and the list is never sold or shared.