Skip to content
AIBRAINWORX / OpenSourceCoder.comLOCAL-FIRST. BUILT IN THE OPEN.
Inference & Quantization / Infrastructure & Data / Inference & Hardware

vLLM v0.34.0 advances the CUDA 13.0 and Model Runner V2 serving line

Published November 16, vLLM v0.34.0 continues the transition away from CUDA 12.9 builds and legacy Model Runner V1.

Listen to this articleBrowser Voice
Ready
GitHub release social card for vllm-project/vllm v0.34.0
AI-generated conceptual illustration by AiBrainWorX; not a photograph or product screenshot. · vLLM project icon; identification only, no endorsement.
Event / context: vLLM v0.34.0 was published November 16, 2026.. Source check: October 28, 2026.

vLLM v0.34.0 landed on November 16, 2026, continuing the v0.33+ serving architecture that standardized on Model Runner V2 and CUDA 13.0 builds (requiring R580 or newer NVIDIA drivers).

Upgrade considerations

Clusters still pinned to CUDA 12.9 remain on the v0.32.0 branch. On supported Blackwell and AMD MI355X deployments, the v0.33/v0.34 runner also enables extensible KV-cache sizing post-warmup by default.

EXPLORE THE SOURCE

GitHub repositories

Original reporting; primary project releases reviewed November 16, 2026. No independent hands-on benchmark claimed.

Follow your curiosity.

Pass the signal.