🎉 Introducing Model Catalog. Control every LLM your team uses. Learn More
🎉 Introducing Model Catalog. Control every LLM your team uses. Learn More
🎉 Introducing Model Catalog. Control every
LLM your team uses. Learn More
Enforce Real Time LLM Behavior
Portkey brings 50+ state-of-the-art AI guardrails on top of our open source Gateway. So that, you can synchronously run Guardrails on your requests and route them with precision.
Enforce Real Time LLM Behavior
Portkey brings 50+ state-of-the-art AI guardrails on top of our open source Gateway. So that, you can synchronously run Guardrails on your requests and route them with precision.
Enforce Real Time LLM Behavior
Portkey brings 50+ state-of-the-art AI guardrails on top of our open source Gateway. So that, you can synchronously run Guardrails on your requests and route them with precision.
> npx @portkey-ai/gateway
> npx @portkey-ai/gateway
> npx @portkey-ai/gateway
Enabling 3000+ leading teams to build the future of GenAI

Enabling 3000+ leading teams to build the future of GenAI

Enabling 3000+ leading teams to build the future of GenAI

Enabling 3000+ leading teams to build the future of GenAI

Enabling 3000+ leading teams to build the future of GenAI




Fast Deterministic Checks
Fast Deterministic Checks
Fast Deterministic Checks
Validate JSON schemas, RegEx patterns, Keyword matches, and much more - live, without adding latency to your requests.
Validate JSON schemas, RegEx patterns, Keyword matches, and much more - live, without adding latency to your requests.
World-Class
Guardrail Partners
World-Class
Guardrail Partners
World-Class
Guardrail Partners
Top guardrails platforms integrate with our gateway, so that you can run your custom Guardrail policies directly on Portkey.
Top guardrails platforms integrate with our gateway, so that you can run your custom Guardrail policies directly on Portkey.






Route With Precision
Route With Precision
Route With Precision
Based on guardrail checks, take precise action on your inputs or outputs. Deny, route to a better model, retry, or just build a rich evals dataset.
Based on guardrail checks, take precise action on your inputs or outputs. Deny, route to a better model, retry, or just build a rich evals dataset.
Latest guides and resources

Portkey & Patronus - Bringing Responsible LLMs in Production
Patronus AI's suite of evaluators are now available on the Portkey Gateway.

Securing your Artificial Intelligence via AI Gateways.
Learn how AI gateways like Portkey with security solutions like Pillar security help to...

Real-time Guardrails vs Batch Evals: Understanding Safety...
In the world of Large Language Model (LLM) applications, ensuring quality, safety...
Latest guides and resources

Portkey & Patronus - Bringing Responsible LLMs in Production
Patronus AI's suite of evaluators are now available on the Portkey Gateway.

Securing your Artificial Intelligence via AI Gateways.
Learn how AI gateways like Portkey with security solutions like Pillar security help to...

Real-time Guardrails vs Batch Evals: Understanding Safety...
In the world of Large Language Model (LLM) applications, ensuring quality, safety...
Latest guides and resources

Portkey & Patronus - Bringing Responsible LLMs in Production
Patronus AI's suite of evaluators are now available on the Portkey Gateway.

Securing your Artificial Intelligence via AI Gateways.
Learn how AI gateways like Portkey with security solutions like Pillar security help to...

Real-time Guardrails vs Batch Evals: Understanding Safety...
In the world of Large Language Model (LLM) applications, ensuring quality, safety...
Frequently Asked Questions
Some questions we get asked the most
Do guardrails add latency to requests?
Only if they’re set to run before the request goes to the model. You can also run guardrails after the request is sent. This avoids any added latency but the guardrail result won’t impact the flow.
Do guardrails add latency to requests?
Only if they’re set to run before the request goes to the model. You can also run guardrails after the request is sent. This avoids any added latency but the guardrail result won’t impact the flow.
Do guardrails add latency to requests?
Only if they’re set to run before the request goes to the model. You can also run guardrails after the request is sent. This avoids any added latency but the guardrail result won’t impact the flow.
2nd FAQs
2nd FAQs
2nd FAQs
What happens if I don’t want to block, just observe?
What happens if I don’t want to block, just observe?
What happens if I don’t want to block, just observe?
Are OpenAI Guardrails only for inputs?
Are OpenAI Guardrails only for inputs?
Are OpenAI Guardrails only for inputs?
What’s the difference between basic and pro guardrails?
What’s the difference between basic and pro guardrails?
What’s the difference between basic and pro guardrails?

Real-time guardrails without slowing down your stack

Real-time guardrails without slowing down your stack

Real-time guardrails without slowing down your stack
Products
© 2025 Portkey, Inc. All rights reserved
HIPAA
COMPLIANT
GDPR
Products
© 2025 Portkey, Inc. All rights reserved
HIPAA
COMPLIANT
GDPR
Products
© 2025 Portkey, Inc. All rights reserved
HIPAA
COMPLIANT
GDPR