Prove the lift · AI model performance evaluation
mistral-large vs llama-3.3-70b
+16-point lift for mistral-large on support-ticket-triage
Test summary
This benchmark evaluates mistral-large against llama-3.3-70b on the support-ticket-triage v2 dataset (6 rows), scored by an LLM-as-judge (llama-3.1-8b, run locally). mistral-large reached a 77% pass rate versus 61% for llama-3.3-70b — a +16-point lift.
- Base model
- llama-3.3-70b
- via meta · 61% pass rate
- Refined model
- mistral-large
- via mistral · 77% pass rate · 3.4s p95
Base pass rate
61%
Refined pass rate
77%
Refined p95
3.4s
Est. tokens saved
9.7K
+0pts
refined beats base on 6 rows
0%
Base model0%
Invoked-refined▲1 improved▼0 regressed●5 unchanged
Per-row results
1Classify: billing dispute vs refund request vs cancellation100% → 100%0
2Route a multi-part ticket to the correct queue100% → 100%0
3Priority-tag an outage report vs a feature question100% → 100%0
4Extract the affected order id from a free-text complaint0% → 100%+100
5Summarize a 12-message thread into a one-line disposition100% → 100%0
6Detect churn-risk sentiment in a renewal thread100% → 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