Grok 4.20

Grok 4

Ranks #20 of 43 on reading real contracts into structured billing data — -6.2 pts vs GPT-5.5.

63.1%

Accuracy

#20 of 43

27%

Hallucination

HIGH-confidence & wrong

$34.0

Cost / 1k contracts

$0.03 each · via OpenRouter

30.2s

Median latency

p90 40.9s

±5.6

Run-to-run σ

3 runs

1.86

Value

acc. pts per $/1k

Tokens per contract

Input20,776

Output3,211

Reasoning0

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →