From Dev.to
vLLM 0.33 Adds a Lossy Speculative Verifier: Why Single-Run Accuracy Tables Cannot Tell You It Is Safe
TL;DR What shipped: vLLM 0.33.0 (PyPI, 2026-11-02) includes FLy, an opt-in speculative decoding verifier that deliberately accepts some draft tokens the target model would have rejected. The PR st...