Prove the lift · AI model performance evaluation

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

+14-point lift for gpt-4o on insurance-extraction-eval

Test summary

This benchmark evaluates gpt-4o against claude-sonnet-4-6 on the insurance-extraction-eval v1 dataset (7 rows), scored by an LLM-as-judge (claude-haiku-4-5). gpt-4o reached a 49% pass rate versus 35% for claude-sonnet-4-6 — a +14-point lift.

Base model
claude-sonnet-4-6
via anthropic · 35% pass rate
Refined model
gpt-4o
via openai · 49% pass rate · 0.9s p95
Base pass rate
35%
Refined pass rate
49%
Refined p95
0.9s
Est. tokens saved
2.6K
+0pts
refined beats base on 7 rows
0%
Base model
0%
Invoked-refined
2 improved0 regressed5 unchanged

Per-row results

1Business-interruption period-of-restoration calculation0%100%+100
2Subrogation waiver interaction with additional-insured status0%0%0
3Named-peril coverage determination under a basic HO form0%100%+100
4Replacement cost vs ACV on a depreciated roof100%100%0
5Pro-rata distribution across layered carriers on a covered loss0%0%0
6Suit-limitation clause: is a late-filed claim contractually barred?0%0%0
7Coinsurance 80% clause recovery on an under-insured building100%100%0
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