Unified Memory

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

vision-003

hard receipthandwriting

The prompt

1 image attached (not shown).

user

The customer wrote a tip and a final total by hand on this restaurant receipt. What final total did they write? Reply with only the number (no units or currency symbols).
What we expected
{
  "answer": 82.7,
  "accept": [],
  "type": "number"
}

What each model answered

LFM2.5-VL 3B

✓ pass · score 100 · 0.47 s

82.70
Scoring detail
{
  "type": "number",
  "parsed": 82.7
}

MiMo-V2.6 Distill 9B

✓ pass · score 100 · 1.49 s

82.70
Scoring detail
{
  "type": "number",
  "parsed": 82.7
}

MiniCPM-V 4.7 1B

✓ pass · score 100 · 1.22 s

82.70
Scoring detail
{
  "type": "number",
  "parsed": 82.7
}

← vision-002All Read images itemsvision-004 →