Prove the lift · AI model performance evaluation

gemini-1.5-flash vs mixtral-8x7b

+11-point lift for gemini-1.5-flash on medical-coding-extract

Test summary

This benchmark evaluates gemini-1.5-flash against mixtral-8x7b on the medical-coding-extract v2 dataset (5 rows), scored by an LLM-as-judge (claude-haiku-4-5, run locally). gemini-1.5-flash reached a 55% pass rate versus 44% for mixtral-8x7b — a +11-point lift.

Base model
mixtral-8x7b
via mistral · 44% pass rate
Refined model
gemini-1.5-flash
via google · 55% pass rate · 1.9s p95
Base pass rate
44%
Refined pass rate
55%
Refined p95
1.9s
Est. tokens saved
9.5K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
3 improved0 regressed2 unchanged

Per-row results

1Pull the procedure code from an operative report0%0%0
2Map a discharge summary to the primary ICD-10 code0%100%+100
3Identify a contraindication from a problem list0%100%+100
4Extract medication + dosage from a clinical note0%0%0
5Determine the correct E/M level from a visit note0%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