Unified Memory

Benchmarks / Which Small AI Model Should You Run on a Mac? We Ran 312 Tests / Assistant actions

assistant-actions-017

hard single

The prompt

Tools offered: create_reminder, add_calendar_event, set_timer, set_alarm, control_light, set_thermostat, lock_door, send_message, play_music, get_weather, add_to_list, navigate_to

user

Set a timer for an hour and a half.
What we expected
{
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": {
          "type": "number",
          "value": 5400,
          "tolerance": 0
        }
      },
      "allow_extra_args": [
        "label"
      ]
    }
  ]
}

What each model answered

MiMo-V2.6 Distill 9B

✗ fail · score 0 · 0.93 s

set_timer({"seconds":900,"label":"an hour and a half"})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 900,
        "label": "an hour and a half"
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 0,
      "args": {
        "seconds": false
      }
    }
  ],
  "extra_calls": 0,
  "base": 0
}

Ornith-1.5 9B

✓ pass · score 100 · 0.80 s

set_timer({"seconds":5400})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 5400
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 1,
      "args": {
        "seconds": true
      }
    }
  ],
  "extra_calls": 0,
  "base": 1
}

Granite 4.2 8B

✗ fail · score 0 · 1.41 s

set_timer({"seconds":540,"label":"hour and a half"})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 540,
        "label": "hour and a half"
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 0,
      "args": {
        "seconds": false
      }
    }
  ],
  "extra_calls": 0,
  "base": 0
}

Spark-X2.5 4B

✗ fail · score 0 · 0.33 s

set_timer({"seconds":90,"label":"timer"})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 90,
        "label": "timer"
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 0,
      "args": {
        "seconds": false
      }
    }
  ],
  "extra_calls": 0,
  "base": 0
}

Mellum2.1 12B-A2.5B

✓ pass · score 100 · 0.51 s

set_timer({"seconds":5400,"label":""})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 5400,
        "label": ""
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 1,
      "args": {
        "seconds": true
      }
    }
  ],
  "extra_calls": 0,
  "base": 1
}

Ling-3.0-tiny (7.9B-A1.3B)

✓ pass · score 100 · 0.30 s

set_timer({"seconds":5400,"label":"hour and a half"})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 5400,
        "label": "hour and a half"
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 1,
      "args": {
        "seconds": true
      }
    }
  ],
  "extra_calls": 0,
  "base": 1
}

MiniCPM5 2B

✗ fail · score 0 · 0.26 s

set_timer({"seconds":90})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 90
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 0,
      "args": {
        "seconds": false
      }
    }
  ],
  "extra_calls": 0,
  "base": 0
}

K2-Horizon 0.9B

✗ fail · score 0 · 0.26 s

set_timer({"seconds":90,"label":"hour and a half"})
Scoring detail
{
  "n_calls": 1,
  "calls": [
    {
      "name": "set_timer",
      "args": {
        "seconds": 90,
        "label": "hour and a half"
      },
      "error": null
    }
  ],
  "per_call": [
    {
      "expected": "set_timer",
      "matched_index": 0,
      "score": 0,
      "args": {
        "seconds": false
      }
    }
  ],
  "extra_calls": 0,
  "base": 0
}

← assistant-actions-016All Assistant actions itemsassistant-actions-018 →