Oracle/GPT-4o

GPT-4o

Healthy

gpt-4o-2024-08-06

Region

us-east-1

Uptime (24h)

99.87%

p50 latency

312ms

p99 latency

1240ms

Error rate

0.12%

Cost / 1k tokens

$0.0050

Latency (24h)

p50p95p99

p99

p95

p50

last 12h (UTC)

Failure breakdown

429Rate limited

71%

500Internal error

18%

timeoutTimeout

11%

SDK usageFull docs →

import { OracleClient } from "@inferenceflow/oracle";

const oracle = new OracleClient({ apiKey: process.env.ORACLE_API_KEY });

// One-shot health check before spending
const health = await oracle.check("openai-gpt4o");

if (health.state === "green") {
  // Safe to call — circuit breaker is closed
  await callProvider("openai-gpt4o", prompt);
} else {
  // Fall back to next healthy provider
  const alt = await oracle.nextHealthy({ exclude: ["openai-gpt4o"] });
  await callProvider(alt.id, prompt);
}