Claude Haiku 5.5
United States benchmark
Serving sensitivity This row ran under the board's forced answer-tool call, which switches Claude's extended thinking off. The same model, re-run with tool_choice: auto, scores 89.2% exact (+9.7) and would rank #10 on the live board; the board keeps the request shape its model card records. Read the note.
Exact match by programall 18 output groups, best first
Bar length is the unweighted exact-match rate on that program's outputs; the right column is the program's share of the headline weight, which follows household dollars, so the bars do not average to the headline.
- Person-level Early Head Start eligibility100.0%3.1%
- Reduced-price school meals eligibility100.0%0.1%
- Free school meals eligibility99.0%0.7%
- TANF99.0%0.3%
- Person-level CHIP eligibility98.3%0.2%
- Local income tax98.0%0.3%
- SSI97.9%2.0%
- Person-level WIC eligibility97.7%0.3%
- Person-level Medicare eligibility97.1%11%
- Self-employment tax96.0%2.1%
- Person-level Medicaid eligibility89.6%30%
- Person-level Head Start eligibility89.5%1.2%
- Federal refundable credits88.8%3.7%
- SNAP86.0%4.1%
- State refundable credits82.7%0.6%
- Payroll tax67.7%16%
- State tax before refundable credits64.4%6.0%
- Federal tax before refundable credits54.8%19%
Table view (exact, within 1%, outputs), hardest first
| Program | Exact | Within 1% | n |
|---|---|---|---|
| Federal tax before refundable credits | 54.8% | 57.1% | 84 |
| State tax before refundable credits | 64.4% | 65.5% | 87 |
| Payroll tax | 67.7% | 70.8% | 96 |
| State refundable credits | 82.7% | 82.7% | 98 |
| SNAP | 86.0% | 86.0% | 93 |
| Federal refundable credits | 88.8% | 88.8% | 98 |
| Person-level Head Start eligibility | 89.5% | 89.5% | 38 |
| Person-level Medicaid eligibility | 89.6% | 89.6% | 173 |
| Self-employment tax | 96.0% | 100.0% | 100 |
| Person-level Medicare eligibility | 97.1% | 97.1% | 172 |
| Person-level WIC eligibility | 97.7% | 97.7% | 177 |
| SSI | 97.9% | 97.9% | 97 |
| Local income tax | 98.0% | 98.0% | 98 |
| Person-level CHIP eligibility | 98.3% | 98.3% | 177 |
| Free school meals eligibility | 99.0% | 99.0% | 100 |
| TANF | 99.0% | 99.0% | 100 |
| Person-level Early Head Start eligibility | 100.0% | 100.0% | 38 |
| Reduced-price school meals eligibility | 100.0% | 100.0% | 100 |
Hardest casesworst misses on positive references
- Payroll tax106% offReference $5,040 · predicted $10,400Inspect household #089
- Payroll tax104% offReference $6,518 · predicted $13,330Inspect household #026
- Federal tax before refundable credits100% offReference $2,906 · predicted $0Inspect household #000
- Payroll tax100% offReference $36,044 · predicted $0Inspect household #005
- State tax before refundable credits100% offReference $40,975 · predicted $0Inspect household #005
- State refundable credits100% offReference $155 · predicted $0Inspect household #007
Scores are from the frozen manuscript snapshot under the AI-alone condition: structured responses over the same household facts (whole-scenario or in output subsets, per the serving-configuration table), no tools, graded against PolicyEngine reference outputs. See the leaderboard and paper for methodology. The frozen US annotations cover 8,001 rows whose legacy threshold score is below 1. That universe contains 7,997 of 7,997 exact-match misses and 4 exact hits. Another 2,138 rows with bounded score below 100 were not selected and have no audit annotation.