Skip to main content
PolicyBench

Claude Opus 4.8

Anthropic · AI alone, no tools
United States rank
#17 of 23
US exact
72.6%
United States

United States benchmark

Exact match
72.6%
Bounded score
83.1%
Parse rate
100.0%
Eligibility flags
884/984

Score by programhardest first

ProgramExactWithin 1%n
Federal tax before refundable credits50.0%54.0%100
State tax before refundable credits57.0%57.0%100
SNAP71.0%71.0%100
Person-level Medicaid eligibility78.0%78.0%177
State refundable credits78.0%79.0%100
Payroll tax81.0%84.0%100
Federal refundable credits82.0%82.0%100
Person-level Head Start eligibility86.8%86.8%38
Person-level CHIP eligibility87.6%87.6%177
Person-level WIC eligibility92.1%92.1%177
Reduced-price school meals eligibility94.0%94.0%100
Person-level Medicare eligibility94.4%94.4%177
Free school meals eligibility97.0%97.0%100
SSI97.0%97.0%100
Person-level Early Head Start eligibility97.4%97.4%38
TANF99.0%99.0%100
Local income tax100.0%100.0%100
Self-employment tax100.0%100.0%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.