GLM-5.1

GLM

Ranks #35 of 43 on reading real contracts into structured billing data — -23.4 pts vs GPT-5.5.

45.9%

Accuracy

#35 of 43

28.1%

Hallucination

HIGH-confidence & wrong

$85.0

Cost / 1k contracts

$0.09 each · via OpenRouter

374.9s

Median latency

p90 1959.3s

±17

Run-to-run σ

3 runs

0.54

Value

acc. pts per $/1k

Tokens per contract

Input21,279

Output12,547

Reasoning9,918

Reliability 94.4% · valid structured output across 18 calls.

Want the full picture? How we score →