Prove the lift · AI model performance evaluation
glm-5.2 vs mistral-large
+9-point lift for glm-5.2 on support-ticket-triage
Test summary
This benchmark evaluates glm-5.2 against mistral-large on the support-ticket-triage v1 dataset (5 rows), scored by an LLM-as-judge (llama-3.1-8b). glm-5.2 reached a 68% pass rate versus 59% for mistral-large — a +9-point lift.
- Base model
- mistral-large
- via mistral · 59% pass rate
- Refined model
- glm-5.2
- via z-ai · 68% pass rate · 3.3s p95
Base pass rate
59%
Refined pass rate
68%
Refined p95
3.3s
Est. tokens saved
3.7K
+0pts
refined beats base on 5 rows
0%
Base model0%
Invoked-refined▲2 improved▼0 regressed●3 unchanged
Per-row results
1Classify: billing dispute vs refund request vs cancellation0% → 100%+100
2Route a multi-part ticket to the correct queue100% → 100%0
3Summarize a 12-message thread into a one-line disposition0% → 100%+100
4Extract the affected order id from a free-text complaint100% → 100%0
5Detect 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