Tool profile
An AI gateway and control plane for managing, observing, and routing production model traffic.
AI gateway
Portkey is an infrastructure layer for teams already serious about shipping LLM features. Instead of being the place where prompts are authored or chat happens, it sits in the request path and helps teams route traffic, manage providers, enforce policy, and observe what production AI systems are actually doing.
That makes it especially relevant once a company is juggling multiple model vendors, retries, fallbacks, and governance requirements. If your problem is operational control over AI traffic rather than simply calling one model API, Portkey is in the right category.
Quick fit
Editorial take
Portkey should be judged against AI gateways and control planes such as Helicone or custom in-house proxy layers, not against chatbot apps. The key editorial question is whether your team has enough real production AI traffic to justify a dedicated routing and governance layer.
What it does well
Primary use cases
Fit notes
Compare with
Pressure-test this pick against nearby alternatives.
Pricing snapshot
Portkey has a Free Forever tier. Production starts around $49/month, includes 100K logs and 100K requests per gateway, and additional 100K requests are about $9. Enterprise is custom.
AgentOps
Free planAgent observability
Observability for AI agents with tracing, debugging, session visibility, and production monitoring.
Closer to agent observability than to model hosting or prompt tooling
