Prove the lift · AI model performance evaluation

claude-opus-4-6 vs claude-sonnet-4-5

-1-point lift for claude-opus-4-6 on medical-coding-extract

Test summary

This benchmark evaluates claude-opus-4-6 against claude-sonnet-4-5 on the medical-coding-extract v3 dataset (5 rows), scored by an LLM-as-judge (claude-haiku-4-5, run locally). claude-opus-4-6 reached a 36% pass rate versus 37% for claude-sonnet-4-5 — a -1-point lift.

Base model
claude-sonnet-4-5
via anthropic · 37% pass rate
Refined model
claude-opus-4-6
via anthropic · 36% pass rate · 0.8s p95
Base pass rate
37%
Refined pass rate
36%
Refined p95
0.8s
Est. tokens saved
534
−0pts
no lift — the base model held its ground
0%
Base model
0%
Invoked-refined
▲1 improved▼0 regressed●4 unchanged

Per-row results

1Pull the procedure code from an operative report0% → 100%+100
2Identify a contraindication from a problem list0% → 0%0
3Determine the correct E/M level from a visit note0% → 0%0
4Extract medication + dosage from a clinical note100% → 100%0
5Map a discharge summary to the primary ICD-10 code100% → 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