Claude Opus 4.8

Claude 4.8

Ranks #19 of 43 on reading real contracts into structured billing data — -6.1 pts vs GPT-5.5.

63.2%

Accuracy

#19 of 43

24.3%

Hallucination

HIGH-confidence & wrong

$268

Cost / 1k contracts

$0.27 each · via OpenRouter

40.6s

Median latency

p90 43.8s

±3.8

Run-to-run σ

3 runs

0.24

Value

acc. pts per $/1k

Tokens per contract

Input34,639

Output3,796

Reasoning0

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →