Prove the lift · AI model performance evaluation

mixtral-8x7b vs gemini-1.5-flash

+11-point lift for mixtral-8x7b on financial-doc-extract

Test summary

This benchmark evaluates mixtral-8x7b against gemini-1.5-flash on the financial-doc-extract v3 dataset (5 rows), scored by an LLM-as-judge (claude-haiku-4-5). mixtral-8x7b reached a 58% pass rate versus 47% for gemini-1.5-flash — a +11-point lift.

Base model
gemini-1.5-flash
via google · 47% pass rate
Refined model
mixtral-8x7b
via mistral · 58% pass rate · 3.0s p95
Base pass rate
47%
Refined pass rate
58%
Refined p95
3.0s
Est. tokens saved
7.6K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
1 improved1 regressed3 unchanged

Per-row results

1Extract net revenue and YoY change from an earnings PDF100%0%-100
2Determine the reporting period from a 10-Q header100%100%0
3Identify a going-concern note in the footnotes0%100%+100
4Extract total debt from a balance sheet0%0%0
5Pull the diluted EPS from an income statement100%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 Invoked
Generated with Invoked · Browse all benchmarks