Prove the lift · AI model performance evaluation

claude-opus-4-6 vs mixtral-8x7b

-10-point lift for claude-opus-4-6 on ecommerce-catalog-norm

Test summary

This benchmark evaluates claude-opus-4-6 against mixtral-8x7b on the ecommerce-catalog-norm v2 dataset (5 rows), scored by an LLM-as-judge (claude-haiku-4-5, run locally). claude-opus-4-6 reached a 15% pass rate versus 25% for mixtral-8x7b — a -10-point lift.

Base model
mixtral-8x7b
via mistral · 25% pass rate
Refined model
claude-opus-4-6
via anthropic · 15% pass rate · 0.6s p95
Base pass rate
25%
Refined pass rate
15%
Refined p95
0.6s
Est. tokens saved
857
0pts
no lift — the base model held its ground
0%
Base model
0%
Invoked-refined
1 improved0 regressed3 unchanged

Per-row results

1Normalize a free-text size into a standard enum100%100%0
2Deduplicate two near-identical SKUs0%0%0
3Classify a listing into the right category node100%100%0
4Extract material + color from a product title0%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 Invoked
Generated with Invoked · Browse all benchmarks