GPT-5.4 Mini

GPT-5.4

Ranks #14 of 43 on reading real contracts into structured billing data — -3.8 pts vs GPT-5.5.

65.5%

Accuracy

#14 of 43

18.8%

Hallucination

HIGH-confidence & wrong

$60.7

Cost / 1k contracts

$0.06 each · via OpenRouter

68.1s

Median latency

p90 103.1s

±3.6

Run-to-run σ

3 runs

1.08

Value

acc. pts per $/1k

Tokens per contract

Input21,576

Output9,900

Reasoning7,229

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →