Narev
  • Twitter
  • Create

  • Evals
    • Models
    • Hardware

  • Catalog

> Hub > Evals > Models > …

loading run

> Hub > Evals > Models > Elementary Math Plain-Number Accuracy

Evaluates elementary arithmetic, fractions, decimals, percentages, money, and simple word problems using plain-number answers.

Scorer: NumericDiff

loading chart
Score vs cost per task
#scoremodel
cost
price
ttfttotal tok
1100%qwen/qwen3.5-9b
$0.00005
$0.10 / $0.15 per 1M
1.43s328.5
2100%gpt-5.6-sol
$0.00029
$5.00 / $30.00 per 1M
0.41s30.2
3100%gemini-3.6-flash
$0.00106
$0.75 / $3.75 per 1M
1.71s158.4