Skip to main content
PolicyBench

Claude Haiku 4.5

Anthropic · AI alone, no tools
United States rank
#19 of 23
US exact
71.7%
United States

United States benchmark

Exact match
71.7%
Bounded score
82.1%
Parse rate
100.0%
Eligibility flags
901/984

Score by programhardest first

ProgramExactWithin 1%n
Federal tax before refundable credits46.0%48.0%100
State tax before refundable credits55.0%56.0%100
Payroll tax65.0%71.0%100
Federal refundable credits75.0%75.0%100
State refundable credits79.0%79.0%100
SNAP80.0%80.0%100
Person-level Medicaid eligibility82.5%82.5%177
Person-level Early Head Start eligibility86.8%86.8%38
Person-level CHIP eligibility90.4%90.4%177
Self-employment tax91.0%94.0%100
Reduced-price school meals eligibility92.0%92.0%100
Person-level WIC eligibility92.7%92.7%177
Person-level Medicare eligibility96.6%96.6%177
SSI97.0%97.0%100
Person-level Head Start eligibility97.4%97.4%38
Free school meals eligibility98.0%98.0%100
Local income tax99.0%99.0%100
TANF99.0%99.0%100

Hardest casesworst misses on positive references

Scores are from the frozen manuscript snapshot under the AI-alone condition: one structured response per household, no tools, graded against PolicyEngine reference outputs. See the leaderboard and paper for methodology.