Benchmarks / Which Small AI Model Should You Run on a Mac? We Ran 312 Tests / Sort and triage

triage-014

hard inbox

The prompt

user

TASK: INBOX

From: Ruth Okafor <ruth.okafor.larkspur@inboxly.com>
To: dana.whitfield@larkspuroutfitters.com
Date: Wed, 21 Oct 2026 13:37 -0600
Subject: quick favor

Dana, are you at your desk? I'm stuck in back-to-back meetings with the board and can't talk on the phone.

I need you to pick up five $200 gift cards (Visa or Amazon) for a client appreciation thing this afternoon. Scratch off the backs and email me photos of the codes as soon as you have them. I'll make sure finance reimburses you today.

Please keep this between us for now, it's a surprise.

Ruth
Sent from my phone

Reply with only a JSON object: {"label": "<urgent_action|reply_needed|fyi|newsletter|spam|phishing>"}

{
  "label": "phishing"
}

What each model answered

{"label": "reply_needed"}

{
  "label": {
    "got": "reply_needed",
    "expected": "phishing",
    "correct": false
  }
}

{"label": "reply_needed"}

{
  "label": {
    "got": "reply_needed",
    "expected": "phishing",
    "correct": false
  }
}

{
  "label": "urgent_action"
}

{
  "label": {
    "got": "urgent_action",
    "expected": "phishing",
    "correct": false
  }
}

{"label": "phishing"}

{
  "label": {
    "got": "phishing",
    "expected": "phishing",
    "correct": true
  }
}

{"label": "phishing"}

{
  "label": {
    "got": "phishing",
    "expected": "phishing",
    "correct": true
  }
}

{"label": "urgent_action"}

{
  "label": {
    "got": "urgent_action",
    "expected": "phishing",
    "correct": false
  }
}

{"label": "phishing"}

{
  "label": {
    "got": "phishing",
    "expected": "phishing",
    "correct": true
  }
}

{"label": "urgent_action"}

{
  "label": {
    "got": "urgent_action",
    "expected": "phishing",
    "correct": false
  }
}

← triage-013All Sort and triage itemstriage-015 →