GPT-5.6 Terra

NewGPT-5.6

Ranks #10 of 35 on reading real contracts into structured billing data, -3.2 pts vs GPT-5.5.

72.8%

Accuracy

#10 of 35

2.4%

Hallucination

HIGH-confidence & wrong

$44.3

Cost / 1k contracts

$0.04 each · via OpenRouter

33.2s

Median latency

p90 40.2s

±0

Run-to-run σ

1 runs

1.64

Value

acc. pts per $/1k

Tokens per contract

Input22,747

Output3,585

Reasoning914

Reliability 100% · valid structured output across 30 calls.

Want the full picture? How we score