Claude Sonnet 5

NewClaude 5

Ranks #13 of 43 on reading real contracts into structured billing data — -3.4 pts vs GPT-5.5.

65.9%

Accuracy

#13 of 43

14%

Hallucination

HIGH-confidence & wrong

$185

Cost / 1k contracts

$0.18 each · via OpenRouter

92.6s

Median latency

p90 173.4s

±1.6

Run-to-run σ

3 runs

0.36

Value

acc. pts per $/1k

Tokens per contract

Input34,733

Output11,523

Reasoning1,722

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →