Prove the lift · AI model performance evaluation

claude-sonnet-4-6 vs gpt-4o

-6-point lift for claude-sonnet-4-6 on contract-clause-qa

Test summary

This benchmark evaluates claude-sonnet-4-6 against gpt-4o on the contract-clause-qa v1 dataset (5 rows), scored by an LLM-as-judge (gpt-4o-mini, run locally). claude-sonnet-4-6 reached a 19% pass rate versus 25% for gpt-4o — a -6-point lift.

Base model
gpt-4o
via openai · 25% pass rate
Refined model
claude-sonnet-4-6
via anthropic · 19% pass rate · 2.0s p95
Base pass rate
25%
Refined pass rate
19%
Refined p95
2.0s
Est. tokens saved
1.2K
−0pts
no lift — the base model held its ground
0%
Base model
0%
Invoked-refined
▲1 improved▼1 regressed●3 unchanged

Per-row results

1Extract the auto-renewal notice window in days100% → 0%-100
2Identify the governing-law clause and jurisdiction100% → 100%0
3Determine indemnification scope and carve-outs100% → 100%0
4Locate the termination-for-convenience terms0% → 0%0
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