Prove the lift · AI model performance evaluation

claude-opus-4-6 vs gemini-1.5-flash

+27-point lift for claude-opus-4-6 on insurance-extraction-eval

Test summary

This benchmark evaluates claude-opus-4-6 against gemini-1.5-flash on the insurance-extraction-eval v1 dataset (5 rows), scored by an LLM-as-judge (gpt-4o-mini, run locally). claude-opus-4-6 reached a 80% pass rate versus 53% for gemini-1.5-flash — a +27-point lift.

Base model
gemini-1.5-flash
via google · 53% pass rate
Refined model
claude-opus-4-6
via anthropic · 80% pass rate · 2.8s p95
Base pass rate
53%
Refined pass rate
80%
Refined p95
2.8s
Est. tokens saved
9.7K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
▲2 improved▼0 regressed●3 unchanged

Per-row results

1Aggregate deductible across a loss sequence under a per-occurrence + annual cap0% → 100%+100
2Replacement cost vs ACV on a depreciated roof100% → 100%0
3Suit-limitation clause: is a late-filed claim contractually barred?0% → 100%+100
4Coinsurance 80% clause recovery on an under-insured building100% → 100%0
5Business-interruption period-of-restoration calculation0% → 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