GPT-5.6 Terra

NewGPT-5.6

Ranks #16 of 43 on reading real contracts into structured billing data — -4.7 pts vs GPT-5.5.

64.6%

Accuracy

#16 of 43

16.3%

Hallucination

HIGH-confidence & wrong

$42.0

Cost / 1k contracts

$0.04 each · via OpenRouter

28.8s

Median latency

p90 37.9s

±0.2

Run-to-run σ

3 runs

1.54

Value

acc. pts per $/1k

Tokens per contract

Input21,576

Output3,409

Reasoning799

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →