vLLM: Releases and Overview
High-throughput LLM serving engine built around PagedAttention. Handles concurrent requests and quantized weights (AWQ / GPTQ / FP8) and is widely used on server-class GPUs.
Category: Inference engines and runtimes. Part of our engine and tool tracker.
At a Glance
| Item | Value |
|---|---|
| Repository | vllm-project/vllm |
| License | Apache-2.0 |
| Main language | Python |
| GitHub stars | 91,980 (as of 2026-09-17) |
| Latest release | v0.33.0rc3 (2026-11-02) |
| Install / docs | official documentation |
The project describes itself as: “A high-throughput and memory-efficient inference and serving engine for LLMs”
Release History
Compiled by Local Model Watch from the project’s GitHub releases. Pre-releases are omitted. Where we wrote an article about a release, it is linked in the last column; smaller releases are tracked here only.
| Version | Released | Release notes | Our article |
|---|---|---|---|
| v0.33.0rc3 | 2026-11-02 | GitHub | — |
| v0.32.0 | 2026-10-18 | GitHub | — |
| v0.31.0 | 2026-10-05 | GitHub | vLLM v0.31.0 Released: Fast Restart and Hardware Optimization |
| v0.30.0 | 2026-09-22 | GitHub | vLLM v0.30.0 Released: Fast Start Weight Caching and New Models |
| v0.29.0 | 2026-09-09 | GitHub | vLLM v0.29.0 Released: Model Runner V2 Default and More |
| v0.28.0 | 2026-08-26 | GitHub | — |
| v0.27.1 | 2026-08-11 | GitHub | — |
| v0.27.0 | 2026-08-11 | GitHub | — |
| v0.26.0 | 2026-07-27 | GitHub | — |
| v0.25.1 | 2026-07-14 | GitHub | — |
| v0.25.0 | 2026-07-12 | GitHub | — |
| v0.24.0 | 2026-06-30 | GitHub | — |
| v0.23.0 | 2026-06-15 | GitHub | — |
| v0.22.1 | 2026-06-05 | GitHub | — |
| v0.22.0 | 2026-05-29 | GitHub | — |
| v0.21.0 | 2026-05-15 | GitHub | — |
| v0.20.2 | 2026-05-10 | GitHub | — |
| v0.20.1 | 2026-05-04 | GitHub | — |
| v0.20.0 | 2026-04-28 | GitHub | — |
| v0.19.1 | 2026-04-18 | GitHub | — |
| v0.19.0 | 2026-04-03 | GitHub | — |
| v0.18.1 | 2026-03-31 | GitHub | — |
| v0.18.0 | 2026-03-21 | GitHub | — |
| v0.17.1 | 2026-03-11 | GitHub | — |
| v0.17.0 | 2026-03-07 | GitHub | — |
| v0.16.0 | 2026-02-26 | GitHub | — |
| v0.15.1 | 2026-02-05 | GitHub | — |
| v0.15.0 | 2026-01-29 | GitHub | — |
| v0.14.1 | 2026-01-25 | GitHub | — |
| v0.14.0 | 2026-01-20 | GitHub | — |
| v0.13.0 | 2025-12-19 | GitHub | — |
| v0.12.0 | 2025-12-03 | GitHub | — |
| v0.11.2 | 2025-11-20 | GitHub | — |
| v0.11.1 | 2025-11-19 | GitHub | — |
| v0.11.0 | 2025-10-03 | GitHub | — |
| v0.10.2 | 2025-09-13 | GitHub | — |
| v0.10.1.1 | 2025-08-21 | GitHub | — |
| v0.10.1 | 2025-08-18 | GitHub | — |
| v0.10.0 | 2025-07-25 | GitHub | — |
| v0.9.2 | 2025-07-08 | GitHub | — |
| v0.9.1 | 2025-06-11 | GitHub | — |
| v0.9.0.1 | 2025-05-31 | GitHub | — |
| v0.9.0 | 2025-05-15 | GitHub | — |
| v0.8.5.post1 | 2025-05-03 | GitHub | — |
| v0.8.5 | 2025-04-29 | GitHub | — |
| v0.8.4 | 2025-04-14 | GitHub | — |
| v0.8.3 | 2025-04-06 | GitHub | — |
| v0.8.2 | 2025-03-24 | GitHub | — |
| v0.8.1 | 2025-03-20 | GitHub | — |
| v0.8.0 | 2025-03-19 | GitHub | — |
| v0.7.3 | 2025-02-21 | GitHub | — |
| v0.7.2 | 2025-02-06 | GitHub | — |
| v0.7.1 | 2025-02-02 | GitHub | — |
| v0.7.0 | 2025-01-27 | GitHub | — |
| v0.6.6.post1 | 2024-12-27 | GitHub | — |
| v0.6.6 | 2024-12-27 | GitHub | — |
| v0.6.5 | 2024-12-18 | GitHub | — |
| v0.6.4.post1 | 2024-11-16 | GitHub | — |
| v0.6.4 | 2024-11-15 | GitHub | — |
| v0.6.3.post1 | 2024-10-18 | GitHub | — |
Articles on Local Model Watch
- 2026-10-05 vLLM v0.31.0 Released: Fast Restart and Hardware Optimization
- 2026-09-22 vLLM v0.30.0 Released: Fast Start Weight Caching and New Models
- 2026-09-09 vLLM v0.29.0 Released: Model Runner V2 Default and More
Last updated 2026-11-02 (JST). Facts above come from the GitHub API; the one-paragraph summary was written by the site’s operator with the help of AI.