LiteLLM Rust Gateway benchmarked: 0.7 ms p99 vs 257.7 ms for Python proxy

Benchmarks of the LiteLLM Rust Gateway beta show the standalone Rust path adding about 0.7 ms at p99 with 21.8 MB peak memory and roughly 2,814 requests per second, versus 257.7 ms p99 and 329.5 MB for the LiteLLM Python v1 proxy. The local-mock test disabled logging, persistence and spend tracking, so it is not a full-feature comparison.

• Rust path: 0.7 ms p99, 21.8 MB memory, 2,814 requests per second
• Python v1 proxy: 257.7 ms p99 and 329.5 MB peak memory
• Test disabled logging, persistence and spend tracking
• Hybrid mode keeps Python for HTTP and routing, Rust for forwarding