Bunker Longevity API Errors: Codes and Fixes - Bunker Longevity

The Bunker Longevity API returns errors as JSON with a stable code you can switch on and a human-readable message. HTTP status codes follow standard REST conventions.

Error shape

{
  "error": {
    "code": "invalid_sequence",
    "message": "Sequence contains characters outside the IUPAC amino acid set",
    "request_id": "req_9e12ab"
  }
}

HTTP status codes

Status	Meaning
`400 Bad Request`	Request body is malformed or fails validation
`401 Unauthorized`	Missing or invalid API key
`403 Forbidden`	Key does not have the required scope
`404 Not Found`	Model, deployment, or job does not exist
`409 Conflict`	Deployment is not in a state that allows the requested operation
`429 Too Many Requests`	Rate limit exceeded
`500 Internal Server Error`	Unexpected server error; safe to retry
`503 Service Unavailable`	Capacity temporarily exhausted; retry with backoff

Error codes

Code	Status	When it happens	How to fix
`invalid_api_key`	401	Key is missing, malformed, or revoked	Check the `Authorization` header and rotate if needed
`insufficient_scope`	403	Key lacks the required scope	Use a key with `inference` or `deploy` scope
`model_not_found`	404	Unknown model name in the URL	Check the model catalog for valid names
`deployment_not_ready`	409	Sent a predict request before deployment reached `ready`	Poll `/status` until the deployment is ready
`invalid_sequence`	400	Amino acid input contains unsupported characters	Restrict to the 20 standard IUPAC codes
`sequence_too_long`	400	Input exceeds the model’s maximum length	Split into smaller windows or use a different model
`invalid_smiles`	400	SMILES string does not parse	Validate SMILES locally with RDKit before sending
`constraint_infeasible`	400	MolGen-7 constraints admit no valid molecule	Loosen constraints and try again
`rate_limit_exceeded`	429	Too many requests in the current window	Back off and retry after `X-RateLimit-Reset`
`quota_exceeded`	429	Monthly quota exhausted	Upgrade your plan or wait for the quota to reset
`capacity_exhausted`	503	No GPU capacity available for the requested instance type	Retry with backoff or pick a different `instance_type`
`internal_error`	500	Unexpected server error	Retry with backoff; contact support if it persists

Retry guidance

Do not retry 400-series errors without changing the request. They are deterministic and will fail again.

Retry 429, 500, and 503 with exponential backoff and jitter.
Include a stable request_id on retries so support can trace them.
Reach out at [email protected] with the request_id for any error you cannot resolve.