Prove the lift · AI model performance evaluation

claude-haiku-4-5 vs mixtral-8x7b

+17-point lift for claude-haiku-4-5 on medical-coding-extract

Test summary

This benchmark evaluates claude-haiku-4-5 against mixtral-8x7b on the medical-coding-extract v2 dataset (5 rows), scored by an LLM-as-judge (llama-3.1-8b). claude-haiku-4-5 reached a 40% pass rate versus 23% for mixtral-8x7b — a +17-point lift.

Base model
mixtral-8x7b
via mistral · 23% pass rate
Refined model
claude-haiku-4-5
via anthropic · 40% pass rate · 3.0s p95
Base pass rate
23%
Refined pass rate
40%
Refined p95
3.0s
Est. tokens saved
4.0K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
1 improved1 regressed3 unchanged

Per-row results

1Map a discharge summary to the primary ICD-10 code100%0%-100
2Determine the correct E/M level from a visit note0%100%+100
3Pull the procedure code from an operative report0%0%0
4Identify a contraindication from a problem list0%0%0
5Extract medication + dosage from a clinical note0%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