Prove the lift · AI model performance evaluation

claude-sonnet-4-6 vs gpt-4o

+13-point lift for claude-sonnet-4-6 on support-ticket-triage

Test summary

This benchmark evaluates claude-sonnet-4-6 against gpt-4o on the support-ticket-triage v2 dataset (6 rows), scored by an LLM-as-judge (gpt-4o-mini, run locally). claude-sonnet-4-6 reached a 66% pass rate versus 53% for gpt-4o — a +13-point lift.

Base model
gpt-4o
via openai · 53% pass rate
Refined model
claude-sonnet-4-6
via anthropic · 66% pass rate · 1.9s p95
Base pass rate
53%
Refined pass rate
66%
Refined p95
1.9s
Est. tokens saved
11.4K
+0pts
refined beats base on 6 rows
0%
Base model
0%
Invoked-refined
▲4 improved▼0 regressed●2 unchanged

Per-row results

1Summarize a 12-message thread into a one-line disposition0% → 0%0
2Priority-tag an outage report vs a feature question0% → 100%+100
3Extract the affected order id from a free-text complaint0% → 100%+100
4Detect churn-risk sentiment in a renewal thread0% → 100%+100
5Route a multi-part ticket to the correct queue100% → 100%0
6Classify: billing dispute vs refund request vs cancellation0% → 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