Prove the lift · AI model performance evaluation

gpt-4o-mini vs gemini-1.5-flash

+18-point lift for gpt-4o-mini on insurance-extraction-eval

Test summary

This benchmark evaluates gpt-4o-mini against gemini-1.5-flash on the insurance-extraction-eval v3 dataset (7 rows), scored by an LLM-as-judge (llama-3.1-8b). gpt-4o-mini reached a 59% pass rate versus 41% for gemini-1.5-flash — a +18-point lift.

Base model
gemini-1.5-flash
via google · 41% pass rate
Refined model
gpt-4o-mini
via openai · 59% pass rate · 2.5s p95
Base pass rate
41%
Refined pass rate
59%
Refined p95
2.5s
Est. tokens saved
10.4K
+0pts
refined beats base on 7 rows
0%
Base model
0%
Invoked-refined
▲5 improved▼0 regressed●2 unchanged

Per-row results

1Aggregate deductible across a loss sequence under a per-occurrence + annual cap0% → 0%0
2Replacement cost vs ACV on a depreciated roof0% → 100%+100
3Named-peril coverage determination under a basic HO form0% → 100%+100
4Coinsurance 80% clause recovery on an under-insured building0% → 100%+100
5Business-interruption period-of-restoration calculation0% → 100%+100
6Pro-rata distribution across layered carriers on a covered loss100% → 100%0
7Subrogation waiver interaction with additional-insured status0% → 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