Mistral Large

NewMistral

Ranks #22 of 43 on reading real contracts into structured billing data — -7.4 pts vs GPT-5.5.

61.9%

Accuracy

#22 of 43

28.8%

Hallucination

HIGH-confidence & wrong

$18.0

Cost / 1k contracts

$0.02 each · via OpenRouter

73.9s

Median latency

p90 98.8s

±5.2

Run-to-run σ

3 runs

3.43

Value

acc. pts per $/1k

Tokens per contract

Input24,045

Output4,000

Reasoning0

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →