Blog

mistral.rs v0.9.3: up to 19.5% higher serving throughput than vLLM

By Eric Buehler | August 28, 2026 | benchmarks, cuda, fp8