vLLM v0.20.0 Ships TurboQuant 2-Bit KV Cache — 4x Capacity with DeepSeek V4 Support

vLLM / GitHub·medium signal

vLLM v0.20.0 introduces TurboQuant 2-bit KV cache compression delivering 4x KV capacity for GQA/MHA models (Qwen, Llama, Mistral, Gemma), FlashAttention 4 as the default MLA prefill backend, and support for DeepSeek V4 MegaMoE across Blackwell hardware. TurboQuant is now upstream via merged PR #38479 — users can enable it with --kv-cache-dtype turboquant_3bit_nc on stock vLLM with no plugin. Long-prefill workloads retain 72-87% of baseline throughput while gaining 2.6-4.9x KV capacity.