Prove the lift · AI model performance evaluation
qwen2.5-vl-32b vs o3-mini
+14-point lift for qwen2.5-vl-32b on financial-doc-extract
Test summary
This benchmark evaluates qwen2.5-vl-32b against o3-mini on the financial-doc-extract v1 dataset (5 rows), scored by an LLM-as-judge (gpt-4o-mini). qwen2.5-vl-32b reached a 66% pass rate versus 52% for o3-mini — a +14-point lift.
- Base model
- o3-mini
- via openai · 52% pass rate
- Refined model
- qwen2.5-vl-32b
- via qwen · 66% pass rate · 1.6s p95
Base pass rate
52%
Refined pass rate
66%
Refined p95
1.6s
Est. tokens saved
10.0K
+0pts
refined beats base on 5 rows
0%
Base model0%
Invoked-refined▲3 improved▼0 regressed●2 unchanged
Per-row results
1Extract net revenue and YoY change from an earnings PDF0% → 100%+100
2Identify a going-concern note in the footnotes0% → 100%+100
3Extract total debt from a balance sheet0% → 100%+100
4Pull the diluted EPS from an income statement100% → 100%0
5Determine the reporting period from a 10-Q header100% → 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