Performance Benchmarks
Embex leverages a high-performance Rust core with SIMD acceleration to deliver speedups over native Python/JS clients.
Python Benchmarks
Section titled “Python Benchmarks”Benchmarks run on M1 Max with 10k vectors (384 dimensions).
| Provider | Client | Insert (ops/s) | Speedup | Search Latency |
|---|---|---|---|---|
| Qdrant | Embex | 24,825 | 4.3x | 1.95ms |
| Native | 5,754 | 4.69ms | ||
| Weaviate | Embex | 5,163 | 4.1x | 1.77ms |
| Native | 1,256 | 4.03ms | ||
| Chroma | Embex | 3,136 | 1.0x | 3.97ms |
| Native | 3,077 | 3.46ms |
(Graphs linked from repository assets)
Micro-Benchmarks
Section titled “Micro-Benchmarks”| Operation | Python | Embex (Rust) | Speedup |
|---|---|---|---|
| Normalization | 45ms | 11ms | 4.1x |
| Cosine Sim | 230ms | 58ms | 4.0x |