Prove the lift · AI model performance evaluation

claude-sonnet-4-6 vs mistral-large

+13-point lift for claude-sonnet-4-6 on insurance-extraction-eval

Test summary

This benchmark evaluates claude-sonnet-4-6 against mistral-large on the insurance-extraction-eval v1 dataset (8 rows), scored by an LLM-as-judge (llama-3.1-8b, run locally). claude-sonnet-4-6 reached a 62% pass rate versus 49% for mistral-large — a +13-point lift.

Base model
mistral-large
via mistral · 49% pass rate
Refined model
claude-sonnet-4-6
via anthropic · 62% pass rate · 2.0s p95
Base pass rate
49%
Refined pass rate
62%
Refined p95
2.0s
Est. tokens saved
3.2K
+0pts
refined beats base on 8 rows
0%
Base model
0%
Invoked-refined
▲4 improved▼2 regressed●2 unchanged

Per-row results

1Suit-limitation clause: is a late-filed claim contractually barred?0% → 100%+100
2Named-peril coverage determination under a basic HO form0% → 100%+100
3Coinsurance 80% clause recovery on an under-insured building100% → 0%-100
4Pro-rata distribution across layered carriers on a covered loss100% → 100%0
5Replacement cost vs ACV on a depreciated roof100% → 0%-100
6Aggregate deductible across a loss sequence under a per-occurrence + annual cap0% → 100%+100
7Business-interruption period-of-restoration calculation0% → 100%+100
8Subrogation waiver interaction with additional-insured status100% → 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