Prove the lift · AI model performance evaluation

claude-opus-4-6 vs llama-3.1-70b

+13-point lift for claude-opus-4-6 on legal-research-qa

Test summary

This benchmark evaluates claude-opus-4-6 against llama-3.1-70b on the legal-research-qa v2 dataset (5 rows), scored by an LLM-as-judge (gpt-4o-mini). claude-opus-4-6 reached a 72% pass rate versus 59% for llama-3.1-70b — a +13-point lift.

Base model
llama-3.1-70b
via ollama · 59% pass rate
Refined model
claude-opus-4-6
via anthropic · 72% pass rate · 1.5s p95
Base pass rate
59%
Refined pass rate
72%
Refined p95
1.5s
Est. tokens saved
10.4K
+0pts
refined beats base on 5 rows
0%
Base model
0%
Invoked-refined
▲1 improved▼1 regressed●2 unchanged

Per-row results

1Summarize the holding of a cited appellate case100% → 0%-100
2Identify the controlling statute in a fact pattern0% → 100%+100
3Determine whether a precedent was overruled100% → 100%0
4Extract the standard of review applied100% → 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