Service Status
FreeGET /api/statusThe /api/status endpoint returns the live operational status of major AI services. Polled every 5 minutes by a Cloudflare Worker cron that hits each provider's status page directly. Status values are normalized to operational, degraded, down, or unknown for easy programmatic consumption.
When to use this endpoint
When your agent needs to know whether a specific AI service is up before routing a request. Pair with /api/premium/routing to automatically avoid down providers.
Example response
{
"ok": true,
"source": "tensorfeed.ai",
"checked": "2026-04-27T18:00:00Z",
"services": [
{
"name": "Anthropic",
"provider": "anthropic",
"status": "operational",
"statusPageUrl": "https://status.anthropic.com",
"components": [{ "name": "API", "status": "operational" }],
"lastChecked": "2026-04-27T17:55:00Z"
}
]
}Code samples
Python SDK
from tensorfeed import TensorFeed
tf = TensorFeed()
status = tf.status()
for s in status["services"]:
if s["status"] != "operational":
print(f"{s['name']} is {s['status']}")TypeScript SDK
import { TensorFeed } from 'tensorfeed';
const tf = new TensorFeed();
const { services } = await tf.status();
const incidents = services.filter(s => s.status !== 'operational');
console.log(`${incidents.length} services with issues`);MCP tool
Available via the TensorFeed MCP server as get_ai_status. Add npx -y @tensorfeed/mcp-server to your Claude Desktop or Claude Code MCP config.
FAQ
How does TensorFeed determine service status?
A Cloudflare Worker cron polls each provider's status page (Atlassian Statuspage JSON or HTML banner parsing) every 5 minutes. We normalize to operational, degraded, down, or unknown for consistent agent consumption.
Which AI services does the status endpoint cover?
Anthropic (Claude), OpenAI (ChatGPT, API), Google (Gemini), Mistral, Cohere, Replicate, HuggingFace, Perplexity, Microsoft Copilot, and Midjourney. The list is in worker/src/sources.ts and grows as we add providers.
How accurate is the status data?
The data is as accurate as each provider's own status page. We do not run independent probes against the actual APIs. If a provider's status page is wrong, our reported status will be wrong too.