GPT-5.4 Nano

GPT-5.4

Ranks #24 of 43 on reading real contracts into structured billing data — -8.9 pts vs GPT-5.5.

60.4%

Accuracy

#24 of 43

14.7%

Hallucination

HIGH-confidence & wrong

$12.8

Cost / 1k contracts

$0.01 each · via OpenRouter

47.7s

Median latency

p90 65.5s

±8

Run-to-run σ

3 runs

4.73

Value

acc. pts per $/1k

Tokens per contract

Input21,576

Output6,773

Reasoning3,971

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →