Prove the lift · AI model performance evaluation
nemotron-ultra-253b vs gemini-3-pro
+7-point lift for nemotron-ultra-253b on contract-clause-qa
Test summary
This benchmark evaluates nemotron-ultra-253b against gemini-3-pro on the contract-clause-qa v3 dataset (5 rows), scored by an LLM-as-judge (llama-3.1-8b). nemotron-ultra-253b reached a 67% pass rate versus 60% for gemini-3-pro — a +7-point lift.
- Base model
- gemini-3-pro
- via google · 60% pass rate
- Refined model
- nemotron-ultra-253b
- via nvidia · 67% pass rate · 3.4s p95
Base pass rate
60%
Refined pass rate
67%
Refined p95
3.4s
Est. tokens saved
1.6K
+0pts
refined beats base on 5 rows
0%
Base model0%
Invoked-refined▲1 improved▼0 regressed●4 unchanged
Per-row results
1Determine indemnification scope and carve-outs100% → 100%0
2Extract the auto-renewal notice window in days100% → 100%0
3Identify the governing-law clause and jurisdiction0% → 100%+100
4Locate the termination-for-convenience terms100% → 100%0
5Find the liability cap and its exceptions100% → 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