Errors

Eridian returns structured errors on every failed request. The body is stable across REST and SDKs. Correlate with logs using error.request_id and X-Eridian-Request-Id (they match).

Error Object

HTTP/1.1 429 Too Many Requests
Content-Type: application/json
X-Eridian-Request-Id: axm_req_8f3a2b1c
X-Eridian-RateLimit-Reset: 1718400060

{
  "error": {
    "type": "eridian.error",
    "code": "rate_limit_exceeded",
    "message": "Project prj_legal_001 exceeded 300 requests per minute.",
    "request_id": "axm_req_8f3a2b1c",
    "doc_url": "https://geteridian.com/docs/rate-limits",
    "param": null
  }
}

object is omitted on errors. SDKs throw EridianError with these fields.

Status Codes

Retry Guidance

Retry only when the table says Yes, and only with exponential backoff plus jitter.

Honor Retry-After and X-Eridian-RateLimit-Reset on 429. Do not retry POST /v1/inference without an Idempotency-Key; a retry without the header can double-spend.

Python

from eridian import Eridian
from eridian.errors import EridianError, RateLimitError

client = Eridian(api_key="eridian_sk_...", max_retries=3)

try:
    client.inference.create(
        model="auto",
        project_id="prj_legal_001",
        messages=[{"role": "user", "content": "Summarize."}],
    )
except RateLimitError as exc:
    print(exc.request_id, exc.code)
except EridianError as exc:
    raise

TypeScript

import { Eridian, EridianError } from "@eridian/sdk";

const client = new Eridian({ apiKey: process.env.ERIDIAN_API_KEY!, maxRetries: 3 });

try {
  await client.inference.create({
    model: "auto",
    projectId: "prj_legal_001",
    messages: [{ role: "user", content: "Summarize." }],
  });
} catch (error) {
  if (error instanceof EridianError) {
    console.error(error.code, error.requestId);
  }
  throw error;
}

See Idempotency and Rate Limits.