Prove the lift · AI model performance evaluation
gemma-3-27b vs minimax-m3
+19-point lift for gemma-3-27b on ecommerce-catalog-norm
Test summary
This benchmark evaluates gemma-3-27b against minimax-m3 on the ecommerce-catalog-norm v2 dataset (5 rows), scored by an LLM-as-judge (gpt-4o-mini). gemma-3-27b reached a 46% pass rate versus 27% for minimax-m3 — a +19-point lift.
- Base model
- minimax-m3
- via minimax · 27% pass rate
- Refined model
- gemma-3-27b
- via google · 46% pass rate · 2.7s p95
Base pass rate
27%
Refined pass rate
46%
Refined p95
2.7s
Est. tokens saved
8.0K
+0pts
refined beats base on 5 rows
0%
Base model0%
Invoked-refined▲1 improved▼0 regressed●3 unchanged
Per-row results
1Classify a listing into the right category node0% → 0%0
2Extract material + color from a product title100% → 100%0
3Deduplicate two near-identical SKUs0% → 100%+100
4Normalize a free-text size into a standard enum100% → 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 InvokedGenerated with Invoked · Browse all benchmarks