Decisions API returns near-uniform probabilities for every question, including the quickstart example

Bug Reports
1

Pre-flight checklist

  • This is about the Perplexity API, not the Perplexity app, Computer, Comet, or web UI.
  • I have removed API keys, secrets, and private data.
  • I have included enough information for someone else to reproduce the issue.

Bug type

  • Incorrect or unexpected model output

Affected API area

  • Decisions API (POST /v1/decisions)

Summary

pplx-decider-v1.1-27b returns near-uniform probabilities for every question we send, including the quickstart example. The same request shapes returned confident, correct answers for us on 2026-10-08.

Expected behavior

For the quickstart review, defect close to 1 and sentiment clearly weighted toward negative.

Actual behavior

sentiment is an even three-way split and defect is 0.73. Every question sits near the uniform prior, e.g. a noul “Is the sky green?” returns 0.41.

Minimal reproduction

The quickstart example, unchanged:

curl -s https://api.perplexity.ai/v1/decisions \
  -H "Authorization: Bearer $PERPLEXITY_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "pplx-decider-v1.1-27b",
    "state": {
      "title": "Battery died after two weeks",
      "review": "The headphones sound great, but the battery stopped charging after two weeks."
    },
    "questions": {
      "defect": {"type": "noul", "instructions": "Does the review report a product defect?"},
      "sentiment": {
        "type": "choice",
        "instructions": "What is the overall sentiment of the review?",
        "criteria": {"positive": null, "mixed": null, "negative": null}
      }
    }
  }'

Response or error

HTTP 200:

{"model":"pplx-decider-v1.1-27b","answers":{"defect":{"type":"noul","noul":0.7316924307587059},"sentiment":{"type":"choice","choice":"negative","confidence":0.003439309703517079,"probabilities":{"positive":0.33112626029180947,"mixed":0.3332475332391792,"negative":0.33562620646901137}}},"usage":{"input_tokens":235,"output_tokens":2}}

Request details

  • Endpoint: POST https://api.perplexity.ai/v1/decisions
  • Model or preset: pplx-decider-v1.1-27b (pplx-decider-v1-27b returns byte-identical numbers)
  • SDK/language/version: raw HTTP (Python httpx)
  • Request ID: ccbfbc6c-3c97-4289-a935-129d8da8af0f, 9ddeaecc-8138-4fe9-88d4-cf4a160073c0
  • Approximate time and timezone: 2026-10-13 around 11:35 UTC; first seen around 10:47 UTC
  • Consistent or intermittent: consistent; repeated calls return identical numbers
2

Thank you for the report. I connected with the engineering team and they’ve made some updates. Can you please try again?

3

Thanks Andrew! Looks like it’s behaving normally again.