Prove the lift · AI model performance evaluation

gpt-5 vs grok-4

+9-point lift for gpt-5 on contract-clause-qa

Test summary

This benchmark evaluates gpt-5 against grok-4 on the contract-clause-qa v1 dataset (5 rows), scored by an LLM-as-judge (gpt-4o-mini, run locally). gpt-5 reached a 51% pass rate versus 42% for grok-4 — a +9-point lift.

Base model
grok-4
via xai · 42% pass rate
Refined model
gpt-5
via openai · 51% pass rate · 0.6s p95
Base pass rate
42%
Refined pass rate
51%
Refined p95
0.6s
Est. tokens saved
7.5K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
2 improved1 regressed2 unchanged

Per-row results

1Locate the termination-for-convenience terms100%0%-100
2Identify the governing-law clause and jurisdiction0%0%0
3Determine indemnification scope and carve-outs0%0%0
4Extract the auto-renewal notice window in days0%100%+100
5Find the liability cap and its exceptions0%100%+100
Prove the lift on your own data.

Run any two models over your tasks, judged automatically, and see the improvement — in seconds.

Try Invoked
Generated with Invoked · Browse all benchmarks