Prove the lift · AI model performance evaluation
gemini-2.5-pro vs grok-4
+12-point lift for gemini-2.5-pro on support-ticket-triage
Test summary
This benchmark evaluates gemini-2.5-pro against grok-4 on the support-ticket-triage v1 dataset (6 rows), scored by an LLM-as-judge (claude-haiku-4-5). gemini-2.5-pro reached a 65% pass rate versus 53% for grok-4 — a +12-point lift.
- Base model
- grok-4
- via xai · 53% pass rate
- Refined model
- gemini-2.5-pro
- via google · 65% pass rate · 2.3s p95
Base pass rate
53%
Refined pass rate
65%
Refined p95
2.3s
Est. tokens saved
8.5K
+0pts
refined beats base on 6 rows
0%
Base model0%
Invoked-refined▲1 improved▼2 regressed●3 unchanged
Per-row results
1Extract the affected order id from a free-text complaint100% → 100%0
2Detect churn-risk sentiment in a renewal thread100% → 0%-100
3Route a multi-part ticket to the correct queue100% → 100%0
4Classify: billing dispute vs refund request vs cancellation0% → 100%+100
5Priority-tag an outage report vs a feature question100% → 0%-100
6Summarize a 12-message thread into a one-line disposition100% → 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