Prove the lift · AI model performance evaluation

mistral-large vs gpt-oss-20b

+18-point lift for mistral-large on insurance-extraction-eval

Test summary

This benchmark evaluates mistral-large against gpt-oss-20b on the insurance-extraction-eval v3 dataset (8 rows), scored by an LLM-as-judge (claude-haiku-4-5). mistral-large reached a 50% pass rate versus 32% for gpt-oss-20b — a +18-point lift.

Base model
gpt-oss-20b
via openai · 32% pass rate
Refined model
mistral-large
via mistral · 50% pass rate · 1.3s p95
Base pass rate
32%
Refined pass rate
50%
Refined p95
1.3s
Est. tokens saved
5.0K
+0pts
refined beats base on 8 rows
0%
Base model
0%
Invoked-refined
▲5 improved▼0 regressed●3 unchanged

Per-row results

1Pro-rata distribution across layered carriers on a covered loss0% → 100%+100
2Suit-limitation clause: is a late-filed claim contractually barred?0% → 100%+100
3Subrogation waiver interaction with additional-insured status0% → 100%+100
4Named-peril coverage determination under a basic HO form0% → 100%+100
5Business-interruption period-of-restoration calculation0% → 100%+100
6Aggregate deductible across a loss sequence under a per-occurrence + annual cap100% → 100%0
7Replacement cost vs ACV on a depreciated roof0% → 0%0
8Coinsurance 80% clause recovery on an under-insured building0% → 0%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