-
Notifications
You must be signed in to change notification settings - Fork 1.4k
Pull requests: vllm-project/vllm-omni
Author
Label
Projects
Milestones
Reviews
Assignee
Sort
Pull requests list
[Perf][MiniCPM-o] Cut Talker per-step host overhead in the sampling path
#5945
opened Aug 9, 2026 by
worryzyy
Loading…
[CI]demote slow nightly cases and trim Qwen-Image-Edit coverage
#5944
opened Aug 9, 2026 by
yenuo26
Collaborator
Loading…
[Kernel] Fix Fish attention fallback on strided KV cache
#5943
opened Aug 9, 2026 by
pxljs
Contributor
Loading…
[Bugfix][NPU] Fix MiniMax-H3 RoPE and padding handling
#5941
opened Aug 9, 2026 by
happyyzy
Loading…
[Perf][MiniCPM-o] Add A3 low-latency single-stream deploy config
#5939
opened Aug 9, 2026 by
worryzyy
Loading…
[Perf][MiniCPM-o] Add configurable first codec chunk threshold to cut TTFP
#5938
opened Aug 9, 2026 by
worryzyy
Loading…
[Perf][Diffusion] MiniMax-H3: leaner VAE decode (autocast cache off, in-place revert, uint8 output)
#5937
opened Aug 9, 2026 by
brandneway
Contributor
•
Draft
[Diffusion] [batch invariance] Install batch invariance in the diffusion worker and document its verified scope [2/2]
#5935
opened Aug 9, 2026 by
Ronnie-Rui
Contributor
•
Draft
[CI]drop DreamZero upstream parity e2e; keep OpenPI serving contract
#5934
opened Aug 9, 2026 by
yenuo26
Collaborator
Loading…
[CI/Build] Add g2pM/g2p-en to dev extras for SoulX SVS weekly CI (#5745)
#5933
opened Aug 9, 2026 by
yenuo26
Collaborator
Loading…
2 tasks
[Diffusion][Perf] Add Qwen-Image QK RoPE Triton path
#5931
opened Aug 8, 2026 by
dongbo910220
Contributor
Loading…
[Perf] MiniMax H3: Add online NVFP4 for the text encoder
#5930
opened Aug 8, 2026 by
nagisa-kunhah
Contributor
•
Draft
[Diffusion] Add component-selective encoder blockwise offload
#5929
opened Aug 8, 2026 by
lishunyang12
Collaborator
•
Draft
[BugFix][Test] Initialize weights in MoT und-mode linear tests
#5928
opened Aug 8, 2026 by
congw729
Collaborator
Loading…
Add Cosmos3-Super serving performance dashboard entry (8x H200)
#5926
opened Aug 8, 2026 by
stewtong
Loading…
[Diffusion] Fuse Qwen-Image-Edit select01 modulation for CUDA
#5921
opened Aug 8, 2026 by
dongbo910220
Contributor
Loading…
[Model] Add native MAGI-2 Preview diffusion support
#5918
opened Aug 8, 2026 by
hsliuustc0106
Collaborator
•
Draft
[Bugfix][StepAudio2] Fix async-chunk metadata mismatch and boundary glitches
#5917
opened Aug 8, 2026 by
sphinxkkkbc
Contributor
Loading…
[Perf] MiniMax H3 Qwen3-VL support fused RMSNorm on NPU
#5915
opened Aug 8, 2026 by
wjialish
Loading…
[Model] Add SkyReels V3 V2V (single-shot video extension, 14B)
diffusion
codes related to diffusion models
new model
add new model
#5913
opened Aug 8, 2026 by
NancyFyong
•
Draft
3 of 7 tasks
[Diffusion] Enable independent requests for rank-local DLO DP
diffusion
codes related to diffusion models
enhancement
New feature or request
#5911
opened Aug 8, 2026 by
lishunyang12
Collaborator
Loading…
[Diffusion][Quantization] Enable MiniMax-H3 global FP8 with DLO
diffusion
codes related to diffusion models
quantization
Code related to quantization
#5910
opened Aug 8, 2026 by
lishunyang12
Collaborator
Loading…
Add measured 8x H200 numbers to the Cosmos3-Super recipe
documentation
Improvements or additions to documentation
#5909
opened Aug 7, 2026 by
stewtong
Loading…
Previous Next
ProTip!
Type g p on any issue or pull request to go back to the pull request listing page.