GitHub
16 articles in the TruthFoundry index. Each links out to the original.
- vLLM releases verified proto-v0.5.0 tag · Developer Bugen Zhao releases a verified proto-v0.5.0 tag for the vLLM project.
- vLLM v0.31.0rc3 Adds Support for Randomized Dummy Inputs · vLLM releases v0.31.0rc3 with new support for randomized dummy inputs.
- vllm-project publishes verified proto-v0.4.0 software release · vllm-project released verified proto-v0.4.0 tagged by committer BugenZhao.
- vLLM v0.30.1rc0 release adds ROCm MI355 kernel support · vLLM project publishes v0.30.1rc0 pre-release with ROCm CI kernel updates.
- vLLM Project Releases Version 0.30.0 With New Models And Optimizations · vLLM publishes v0.30.0 release with new model support, performance gains, and breaking changes.
- vLLM releases proto-v0.3.0 protocol update · Developer BugenZhao releases vLLM proto-v0.3.0 protocol version.
- vLLM releases verified proto-v0.2.0 tag · Developer BugenZhao released a verified vLLM proto-v0.2.0 tag via GitHub.
- vLLM Releases v0.29.1rc0 with Dual-Key Gumbel-Max Watermarking · vLLM releases version 0.29.1rc0 introducing dual-key gumbel-max watermarking for speculative decoding.
- vllm-project publishes proto-v0.1.0 tagged software release · vllm-project has released the proto-v0.1.0 tagged software version.
- vLLM 0.29.0 Releases Model Runner V2, New Models, and Breaking Changes · vLLM 0.29.0 defaults Model Runner V2 and adds support for new models like Hy4-preview and Qwen3.8.
- vLLM project publishes v0.29.0rc6 pre-release on GitHub · vllm-project tagged v0.29.0rc6 release candidate with core bugfix on GitHub.
- vLLM releases v0.29.0rc5 with Mamba prefix cache retention change · vLLM project released version v0.29.0rc5 on 2025-09-08, including a Mamba prefix cache retention change.
- vLLM releases v0.28.1rc0 with tool and recipe improvements · vLLM project tagged v0.28.1rc0 release on GitHub, improving sweep recommendations and short-alias parsing.
- vLLM 0.28.0 Release: Kimi-K3 Optimizations, DeepSeek V4 Support, and Model Runner V2 · vLLM 0.28.0 introduces major optimizations for Kimi-K3 and DeepSeek V4 alongside Model Runner V2 maturation.
- vLLM releases v0.27.1 patch adding DSpark Markov head support · vLLM project releases v0.27.1 patch update to support quantized DSpark Markov heads.
- vLLM v0.27.0 Adds Kimi K3, Qwen3.5, FlashAttention 4, and DeepSeek-V4 Optimizations · vLLM v0.27.0 introduces Kimi K3 support, FlashAttention 4, and significant performance gains for DeepSeek-V4.