Prove the lift · AI model performance evaluation

gpt-4-turbo vs claude-sonnet-4-6

+7-point lift for gpt-4-turbo on financial-doc-extract

Test summary

This benchmark evaluates gpt-4-turbo against claude-sonnet-4-6 on the financial-doc-extract v3 dataset (5 rows), scored by an LLM-as-judge (claude-haiku-4-5). gpt-4-turbo reached a 44% pass rate versus 37% for claude-sonnet-4-6 — a +7-point lift.

Base model
claude-sonnet-4-6
via anthropic · 37% pass rate
Refined model
gpt-4-turbo
via openai · 44% pass rate · 1.0s p95
Base pass rate
37%
Refined pass rate
44%
Refined p95
1.0s
Est. tokens saved
9.2K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
0 improved0 regressed5 unchanged

Per-row results

1Extract net revenue and YoY change from an earnings PDF100%100%0
2Determine the reporting period from a 10-Q header0%0%0
3Pull the diluted EPS from an income statement0%0%0
4Identify a going-concern note in the footnotes0%0%0
5Extract total debt from a balance sheet0%0%0
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