Prove the lift · AI model performance evaluation
gemini-1.5-pro vs claude-haiku-4-5
+21-point lift for gemini-1.5-pro on ecommerce-catalog-norm
Test summary
This benchmark evaluates gemini-1.5-pro against claude-haiku-4-5 on the ecommerce-catalog-norm v3 dataset (5 rows), scored by an LLM-as-judge (llama-3.1-8b). gemini-1.5-pro reached a 62% pass rate versus 41% for claude-haiku-4-5 — a +21-point lift.
- Base model
- claude-haiku-4-5
- via anthropic · 41% pass rate
- Refined model
- gemini-1.5-pro
- via google · 62% pass rate · 1.7s p95
Base pass rate
41%
Refined pass rate
62%
Refined p95
1.7s
Est. tokens saved
11.1K
+0pts
refined beats base on 5 rows
0%
Base model0%
Invoked-refined▲2 improved▼1 regressed●1 unchanged
Per-row results
1Classify a listing into the right category node0% → 100%+100
2Extract material + color from a product title100% → 0%-100
3Normalize a free-text size into a standard enum0% → 0%0
4Deduplicate two near-identical SKUs0% → 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 InvokedGenerated with Invoked · Browse all benchmarks