BENCHMARK
FastAPI
FastAPI: 53 files, 20 real questions, reproducible
Full file baseline
83,681 tok/q
83,681 tok/q
After retrieval
4,927 tok/q
4,927 tok/q
After compression
523 tok/q
523 tok/q
Recall@10
0.90
0.90
94%
retrieval savings
No cherry-picking. No synthetic queries.
20 questions a developer would actually ask.
20 questions a developer would actually ask.
$ python benchmarks/run_benchmark.py