Prove the lift · AI model performance evaluation

claude-3-7-sonnet vs gpt-4o

+11-point lift for claude-3-7-sonnet on code-review-bench

Test summary

This benchmark evaluates claude-3-7-sonnet against gpt-4o on the code-review-bench v2 dataset (6 rows), scored by an LLM-as-judge (claude-haiku-4-5). claude-3-7-sonnet reached a 63% pass rate versus 52% for gpt-4o — a +11-point lift.

Base model
gpt-4o
via openai · 52% pass rate
Refined model
claude-3-7-sonnet
via anthropic · 63% pass rate · 0.6s p95
Base pass rate
52%
Refined pass rate
63%
Refined p95
0.6s
Est. tokens saved
7.5K
+0pts
refined beats base on 6 rows
0%
Base model
0%
Invoked-refined
2 improved0 regressed4 unchanged

Per-row results

1Flag an off-by-one in a pagination loop0%100%+100
2Find a resource leak on an early-return path100%100%0
3Note a broken null-check after a refactor100%100%0
4Detect a missing await on an async DB write0%100%+100
5Catch an unsanitized input reaching a SQL string100%100%0
6Identify an N+1 query in an ORM call site100%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