API reference · CAIN Eval

Scan Contamination

POST /fabric/eval/contamination/scan

Last reviewed 2026-10-02

POST /fabric/eval/contamination/scan

Scan for benchmark contamination using membership-inference style probing.

Compares benchmark item overlap vs control set overlap. High z-score = potential contamination.

Request body

ContaminationScanRequest (required)

Field	Type	Meaning
`model_endpoint` required	string
`benchmark_items` required	array of `BenchmarkItem`
`control_items` required	array of `BenchmarkItem`
`timeout_seconds`	integer	min `1.0` · max `60.0` · default `30`

Responses

Status	Meaning	Body
`200`	Successful Response	JSON
`422`	Validation Error	`HTTPValidationError`
`401`	No key, or a key that is not valid	JSON `detail`
`429`	Rate limit for your plan reached; retry after the Retry-After seconds	JSON `detail`

Try it

Your API key X-API-Key Remember the key in this tab onlyRequest body application/json

This call changes data in the account the key belongs to.

The request goes from your browser straight to this site. Your key is not stored unless you tick the box, and then only in this tab's session storage.

Code

curl -sS -X POST 'https://cainstudio.online/fabric/eval/contamination/scan' \
  -H "X-API-Key: $CAIN_API_KEY" \
  -H 'Content-Type: application/json' \
  -d '{"model_endpoint": "model-endpoint", "benchmark_items": [{"prompt_prefix": "prompt-prefix", "canonical_continuation": "canonical-continuation"}], "control_items": [{"prompt_prefix": "prompt-prefix", "canonical_continuation": "canonical-continuation"}], "timeout_seconds": 30}'

import json, os, urllib.request

req = urllib.request.Request('https://cainstudio.online/fabric/eval/contamination/scan', method='POST',
    headers={"X-API-Key": os.environ["CAIN_API_KEY"], "Content-Type": "application/json"},
    data=json.dumps({
      "model_endpoint": "model-endpoint",
      "benchmark_items": [
        {
          "prompt_prefix": "prompt-prefix",
          "canonical_continuation": "canonical-continuation"
        }
      ],
      "control_items": [
        {
          "prompt_prefix": "prompt-prefix",
          "canonical_continuation": "canonical-continuation"
        }
      ],
      "timeout_seconds": 30
    }).encode())
with urllib.request.urlopen(req, timeout=60) as r:
    print(r.status, json.load(r))

const res = await fetch("https://cainstudio.online/fabric/eval/contamination/scan", {
  method: "POST",
  headers: {
    "X-API-Key": process.env.CAIN_API_KEY!,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({
    "model_endpoint": "model-endpoint",
    "benchmark_items": [
      {
        "prompt_prefix": "prompt-prefix",
        "canonical_continuation": "canonical-continuation"
      }
    ],
    "control_items": [
      {
        "prompt_prefix": "prompt-prefix",
        "canonical_continuation": "canonical-continuation"
      }
    ],
    "timeout_seconds": 30
  }),
});
console.log(res.status, await res.json());

Schemas used

BenchmarkItem · ContaminationScanRequest · HTTPValidationError · ValidationError

all 7 CAIN Eval calls · Get Scan →

Try CAIN-42 on your own agents

Create a free account and every new account starts with a 7-day trial of the full platform. Or try the sandbox first, with no account at all.

Create a free account → · Try the sandbox · See the whole ecosystem