Is Portkey Down? How to Check Portkey AI Gateway Status in 2026

Portkey has become the go-to AI gateway for teams that need unified routing, observability, and fallback logic across multiple LLM providers. It sits between your application and providers like OpenAI, Anthropic, Google, and Mistral — which means when Portkey goes down, it can silently break your entire AI stack.

Whether you're seeing 502 Bad Gateway errors, requests timing out mid-stream, or observability logs suddenly going silent, this guide will help you determine: is Portkey down, or is the problem upstream with your AI provider?

How to Check if Portkey is Down (Fastest Methods)

1. Check the Official Portkey Status Page

Portkey maintains a live status page at status.portkey.ai. It shows real-time uptime for the gateway API, routing engine, and the observability dashboard. Active incidents are posted here as soon as Portkey's engineering team confirms them.

2. Test the Gateway Endpoint Directly

Send a minimal request to isolate whether the problem is the Portkey gateway or your upstream provider:

curl https://api.portkey.ai/v1/chat/completions \
  -H "x-portkey-api-key: $PORTKEY_API_KEY" \
  -H "x-portkey-provider: openai" \
  -H "Authorization: Bearer $OPENAI_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-4o-mini","messages":[{"role":"user","content":"ping"}]}'

A 401 means an API key problem. A 502 or 504 points to the Portkey gateway itself failing. A 200 from a direct OpenAI call (bypassing Portkey) confirms Portkey is the issue.

3. Bypass Portkey and Call Your Provider Directly

The fastest diagnostic: call OpenAI or Anthropic directly, bypassing the Portkey gateway. If the direct call succeeds and the Portkey-routed call fails, Portkey is down. This also serves as your immediate workaround to keep production running.

4. Search X (Twitter) for Real-Time Developer Reports

Search "Portkey down" or "Portkey AI gateway" on X filtered by Latest. The developer community surfaces AI tool outages on X within minutes — often before the official status page is updated.

5. Use API Status Check for Automated Gateway Monitoring

For production systems, API Status Check monitors Portkey's gateway endpoints every 30 seconds and sends instant alerts via Slack, email, or PagerDuty. You'll know about outages before your users do.

Why Does Portkey Go Down?

Portkey's role as a middleware gateway creates specific failure patterns distinct from direct AI provider outages:

• Routing Engine Failures: Portkey's load balancing and fallback routing logic runs on its own infrastructure. A bug in a routing rule or a routing engine crash can silently drop requests even when all upstream providers are healthy.
• Upstream Provider Cascades: When OpenAI or Anthropic slows down, Portkey's retry and fallback logic can overload its own gateway — turning a partial upstream slowdown into a full gateway outage for high-volume customers.
• Rate Limit Accounting Delays: Portkey tracks token usage across providers for rate limiting and cost tracking. Database lag in the accounting layer can cause requests to be incorrectly throttled or dropped.
• Observability Pipeline Backup: Portkey's logging and analytics pipeline can back up under heavy load, slowing the main request path as logs are written synchronously in some configurations.
• Config Propagation Lag: Changes to virtual keys, provider configs, or routing rules may take time to propagate across Portkey's edge nodes, causing inconsistent behavior that looks like an intermittent outage.
• Planned Maintenance: Portkey schedules maintenance windows for major updates. Check their status page and changelog for scheduled downtime notices.

Common Portkey Error Codes and What They Mean

What to Do When Portkey Is Down

1. Confirm it's Portkey, not your provider: Call your AI provider (OpenAI, Anthropic, etc.) directly, bypassing Portkey entirely. If the direct call works, Portkey is the issue.
2. Switch to direct provider routing temporarily: Update your application's base URL from Portkey's gateway endpoint to the provider's native endpoint. This keeps production running while Portkey resolves the incident.
3. Check your virtual key config: Verify your Portkey virtual keys are active and have remaining budget. Expired or overloaded virtual keys can mimic gateway outages.
4. Monitor the Portkey status page: Subscribe to incident updates at status.portkey.ai for real-time notifications on resolution progress.
5. Set up automated monitoring: Configure API Status Check to ping both your Portkey gateway endpoint and your direct provider endpoint. If Portkey goes down in the future, you'll get an alert within 30 seconds.

Portkey Alternatives When the Gateway is Down

If Portkey is down and you need an AI gateway solution right now, these alternatives can handle routing and load balancing across multiple AI providers:

• LiteLLM: Open-source AI gateway supporting 100+ LLM providers. Self-hostable and battle-tested for production routing and fallbacks.
• Helicone: AI observability and gateway platform with built-in request logging, cost tracking, and provider routing.
• Direct Provider SDK: Fall back to calling OpenAI, Anthropic, or your preferred provider directly using their official SDK — no gateway dependency.
• OpenRouter: Hosted AI routing service with access to 200+ models, built-in fallbacks, and a single unified API.

Frequently Asked Questions