Kaelum Bench

100-task benchmark

Accounting Capstone

FMB-CAP-ACC-03 · Public calibration

Integrated accounting analysis across a multi-step financial case.

What this benchmark tests

Published benchmark scope without claiming separate competency scores.

  • Statement reconstruction across balance sheet, income statement, equity, and cash flow
  • HGB-to-IFRS separation and bridge
  • Accounting adjustments before ratio analysis
  • Earnings quality, leverage, and stress resilience

Benchmark shape

Only definition and published scoring properties are shown.

Difficulty
10 basic · 40 intermediate · 50 advanced
Comparison
Configuration-specific
Available points
1000 available points
AI judge
openai/gpt-5.6-sol
Deterministic component
800 points
AI rubric component
200 points

Published model runs

All 14 models

Select a row to inspect its benchmark-specific evidence.

  • Recorded cost & tokens
    Cost
    $4.81
    Tokens
    926,788
  • Recorded cost & tokens
    Cost
    $5.12
    Tokens
    542,042
  • Recorded cost & tokens
    Cost
    $3.70
    Tokens
    412,131
  • Recorded cost & tokens
    Cost
    $4.36
    Tokens
    492,087
  • Recorded cost & tokens
    Cost
    $4.58
    Tokens
    571,336
  • Recorded cost & tokens
    Cost
    $7.81
    Tokens
    950,805
  • Recorded cost & tokens
    Cost
    $3.13
    Tokens
    819,619
  • Recorded cost & tokens
    Cost
    $3.57
    Tokens
    468,465
  • Recorded cost & tokens
    Cost
    $3.47
    Tokens
    758,625
  • Recorded cost & tokens
    Cost
    $3.42
    Tokens
    431,483
  • Recorded cost & tokens
    Cost
    $3.58
    Tokens
    408,393
  • Recorded cost & tokens
    Cost
    $3.36
    Tokens
    390,685
  • Recorded cost & tokens
    Cost
    $3.65
    Tokens
    435,697
  • Recorded cost & tokens
    Cost
    $3.41
    Tokens
    422,844

Campaign evidence summary

Completed runs
14
Total recorded cost
$57.96
Total tokens
8,031,000
Valid outputs
1378 / 1400