Search results

[Feature]: GRPC interface for Pooling

feature requestNew feature or requestNew feature or request

Status: Open.

#43207 In vllm-project/vllm;

· unix1986 opened on May 20, 2026

[Bug]: Port collision ([Errno 98] Address already in use) when launching multiple LLM(tensor_parallel_size=2) instances concurrently on a single node (V1 Engine)

bugSomething isn't workingSomething isn't working

Status: Open.

#43206 In vllm-project/vllm;

· ycsh15151 opened on May 20, 2026

[Bug]: Not support role tool of image

bugSomething isn't workingSomething isn't working

Status: Open.

#43203 In vllm-project/vllm;

· luojiaaoo opened on May 20, 2026

[Usage]: DeepSeek-V4 startup takes ~8 min — guidance on reducing initialization time

usageHow to use vllmHow to use vllm

Status: Open.

#43189 In vllm-project/vllm;

· tianlang-wq opened on May 20, 2026

[Performance]: Triton fusion for Qwen2/3-MoE shared-expert gate (Qwen2MoeMLP/Qwen3MoeMLP)

performancePerformance-related issuesPerformance-related issues

rocmRelated to AMD ROCmRelated to AMD ROCm

Status: Open.

#43187 In vllm-project/vllm;

· haofrank opened on May 20, 2026

[Feature]: Port Gemma4 vision encoder to MMEncoderAttention with FlashAttention support

feature requestNew feature or requestNew feature or request

Status: Open.

#43178 In vllm-project/vllm;

· lucianommartins opened on May 20, 2026

[Bug]: ZeroDivisionError in deepgemm_post_process_fp8_weight_block when loading FP8 model with TP=16 on dual-node H20

bugSomething isn't workingSomething isn't working

Status: Open.

#43174 In vllm-project/vllm;

· tianlang-wq opened on May 20, 2026

[Usage]: GLM-5/GLM-5.1 deployment on L40 with vLLM

usageHow to use vllmHow to use vllm

Status: Open.

#43172 In vllm-project/vllm;

· zhanghx0905 opened on May 20, 2026

[Feature]: Add separate_reasoning option for OpenAI chat completion API

feature requestNew feature or requestNew feature or request

Status: Open.

#43171 In vllm-project/vllm;

· F0rtiter opened on May 20, 2026

[Bug]: GLM-5.1-FP8 produces gibberish with RunAI streamer after ac3dac545

Status: Open.

#43163 In vllm-project/vllm;

· bbartels opened on May 19, 2026

[Bug][Perf Regression]: AMD MI355X Kimi K2.5/2.6 arch 38% perf regression

bugSomething isn't workingSomething isn't working

rocmRelated to AMD ROCmRelated to AMD ROCm

Status: Open.

#43153 In vllm-project/vllm;

· functionstackx opened on May 19, 2026

[Bug]: UBatch CUDA graph capture stores graph under first-two-microbatch token count when ubatch_size > 2

Status: Open.

#43145 In vllm-project/vllm;

· glaziermag opened on May 19, 2026