Mistral Small

Mistral

Ranks #32 of 43 on reading real contracts into structured billing data — -16.8 pts vs GPT-5.5.

52.5%

Accuracy

#32 of 43

42.9%

Hallucination

HIGH-confidence & wrong

$5.50

Cost / 1k contracts

$0.01 each · via OpenRouter

17.1s

Median latency

p90 18.5s

±3.5

Run-to-run σ

3 runs

9.61

Value

acc. pts per $/1k

Tokens per contract

Input24,045

Output3,091

Reasoning0

Reliability 100% · valid structured output across 18 calls.

Want the full picture? How we score →