Search results

[Bug]: FlashInfer decode + paged fp8 kv 精度问题导致测试被跳过（test_flashinfer.py）

bugSomething isn't workingSomething isn't working

Status: Open.

#315 In MetaX-MACA/vLLM-metax;

· zhanma666 opened on Jul 2, 2026

[Bug]: DeepSeek-V4/MiMo torch.compile hotfixes 缺少自动化版本清理机制"

bugSomething isn't workingSomething isn't working

Status: Open.

#314 In MetaX-MACA/vLLM-metax;

· zhanma666 opened on Jul 2, 2026

[Performance]: MetaX C550 + Qwen3-ASR-1.7B 高并发场景下 vLLM 调度性能严重劣化

Status: Open.

#299 In MetaX-MACA/vLLM-metax;

· txb0318 opened on Jun 29, 2026

[Performance]: N260 缺少 Qwen3.5-122B MoE 的 fused MoE tuning config

Status: Open.

#295 In MetaX-MACA/vLLM-metax;

· qdym188 opened on Jun 22, 2026

[Bug]: MoE topk_softmax/topk_sigmoid return wrong routing weights at (num_experts=128, topk=8)

Status: Open.

#270 In MetaX-MACA/vLLM-metax;

· sxvvv opened on Jun 6, 2026

[Bug] --quantization gguf fails: ggml ops compiled but not registered (torch.ops._C.ggml_dequantize missing)

Status: Open.

#262 In MetaX-MACA/vLLM-metax;

· linkeLi0421 opened on May 30, 2026

[Bug]: qwen 3.6 MTP 输出内容损坏

bugSomething isn't workingSomething isn't working

Status: Open.

#261 In MetaX-MACA/vLLM-metax;

· everything411 opened on May 28, 2026

[Usage]: Does MetaX currently support vllm-omni?

Status: Open.

#259 In MetaX-MACA/vLLM-metax;

· ZhongsJie opened on May 20, 2026

[Feature]: vLLM-metax 测试移植追踪（对比 upstream vllm）

Status: Open.

#254 In MetaX-MACA/vLLM-metax;

· Dayuxiaoshui opened on May 1, 2026

[Bug] Triton kernel might fail on triton-based attention backend

wontfixThis will not be worked onThis will not be worked on

Status: Open.

#252 In MetaX-MACA/vLLM-metax;

· ILikeIneine opened on Apr 28, 2026

[26Q2] Roadmap

Status: Open.

#233 In MetaX-MACA/vLLM-metax;

· ILikeIneine opened on Apr 12, 2026

[MRV2] Fully support MRV2

enhancementNew feature or requestNew feature or request

Status: Open.

Feature

#231 In MetaX-MACA/vLLM-metax;

· ILikeIneine opened on Apr 2, 2026