This website requires JavaScript.
b6d548afd7
[Fix] Resolve VLM test image placeholders from the model's own chat template (#33509 )
Liangsheng Yin
2026-08-04 02:01:39 -07:00
26a542722f
[CI] Replace the rust-ext-build venv instead of failing when it exists (#33512 )
2026-08-04 01:56:16 -07:00
eaf5c29cc5
[AMD] Enable block-fp8 + quick INT4 all-reduce in MiniMax-M3 MI35x nightly Test (#33402 )
YC Yen-Ching Tseng
2026-08-04 16:55:06 +08:00
53804d609c
[CI][XPU] Stabilize XPU CI: pin UMD/IGC, retry infra flakes, right-size EAGLE3 (#32438 )
ashwini rathi and Ma Mingfei
2026-08-04 13:58:52 +05:30
17d19081d9
[mm] sglang-mm: server vision pipeline core (fetch/driver/pipeline) + Qwen VL (#32364 )
Kan Wu and Claude Fable 5
2026-08-04 00:47:00 -07:00
154f0ac662
Fix DSpark and DP/EP (#33098 )
Vladislav Nosivskoy and Xinyuan Tong
2026-08-04 10:35:57 +03:00
bfa4e4a57b
[Nemotron] Hoist mamba track-mask host syncs out of the per-layer prefill path (#32589 )
Alex Nails and Claude Opus 5
2026-08-03 23:58:28 -07:00
23ea7b6481
Prewarm DSV4 MHC post kernel at model load (#30741 )
weireweire and weireweire
2026-08-04 14:41:56 +08:00
157401f050
[CI] Build the Rust extensions on the 5090 pool and seed the cache from main (#33460 )
Liangsheng Yin
2026-08-03 23:33:20 -07:00
101bb2327c
[diffusion] fix: fix local-path detection for MiniMax-H3 and other non-diffusers models (#33365 )
TobyMint and Mick
2026-08-04 14:03:35 +08:00
4494fb96b2
refactor the tcp listener binding logic (#33420 )
Rain Jiang
2026-08-03 22:10:52 -07:00
48dcadc770
[AMD][DI][CI] 5/N Add DSV4 wide-EP16 4-node 2P1D nightly recipes (#31500 )
Zhaoyi Li and Chen
2026-08-04 00:08:00 -05:00
afc868517b
[Perf] Speed up the Kimi-K2.5 vision path and match PIL bicubic in the GPU resize (#33349 )
Liangsheng Yin and Mick
2026-08-03 21:57:49 -07:00
7ba393dd15
[CI] Dispatch base-a-test-cpu through its own reusable stage workflow (#33461 )
Liangsheng Yin
2026-08-03 20:49:23 -07:00
c6f2a9c1d4
[diffusion] Restrict request-level quality to two validated tiers: lossless (default) and high (#33453 )
Xiaoyu Zhang and Claude Fable 5
2026-08-04 11:43:16 +08:00
614825fd38
[vla] fix: pi05 models does not apply scale factor for language embeddings (#33367 )
Jinchen Han and Mick
2026-08-04 11:25:54 +08:00
b058dc9106
[diffusion] fix: reject ring parallelism where it would silently miscompute (#33353 )
Mick
2026-08-04 11:15:23 +08:00
7dd8a3d5ea
[CI] Fix scheduler max new tokens test fixture (#33467 )
Mohammad Miadh Angkad
2026-08-04 11:12:21 +08:00
1e64fc1563
[Fix] Honor FlashMLA natural-log LSE in DCP reduction (#33065 )
EchO
2026-08-04 10:55:52 +08:00
91fae8a72c
[DCP] Bound a request by the aggregate KV pool, not one rank's share (#33448 )
Khoa Pham
2026-08-03 19:36:25 -07:00
c113ead98a
Bump helion version to 1.4 (#32562 )
Oguz Ulgen and Cheng Wan
2026-08-03 19:10:31 -07:00
92087ef4d2
fix(mem_cache): state the MLA KV bound in the DCP index space (#33432 )
Yuwei An and Claude Opus 5
2026-08-03 19:07:49 -07:00
eb31a53338
Revert "Add flashinfer rmsnorm + quant fusion support SM90, SM100, SM120" (#33455 )
Baizhou Zhang
2026-08-03 19:03:56 -07:00
cdff33d738
[CI] Build the Rust extension modules once per run instead of in every CUDA job (#33384 )
Liangsheng Yin
2026-08-03 18:56:48 -07:00
572924634b
[mem_cache] Build empty-prefix last_loc sentinel on-device to avoid per-call H2D sync (#32575 )
Xingyu Liu
2026-08-03 18:55:03 -07:00
a84e70eb1e
[AMD][DI][CI] 6/N Add Kimi-K2.6 MXFP4 wide-EP16 2P1D nightly recipes (#33333 )
Zhaoyi Li
2026-08-03 20:51:55 -05:00
03f44c978a
[AMD][DI][CI] Auto-mount latest host ionic/ibverbs userspace in the MI355X nightly launcher (fix ABI mismatch) (#33374 )
Zhaoyi Li and Chen
2026-08-03 20:14:14 -05:00
7f6a2e2b50
[Refactor] Clean up and split DSA indexer (#33443 )
Baizhou Zhang
2026-08-03 18:05:01 -07:00
3960983753
Add flashinfer rmsnorm + quant fusion support SM90, SM100, SM120 (#32994 )
2026-08-03 17:37:42 -07:00
f829fb3ff7
[diffusion] Fix component accuracy topology reuse (#33317 )
Xiaoyu Zhang
2026-08-04 08:35:03 +08:00
1307968605
[JIT] Drop redundant per-kernel arch overrides (#32952 )
2026-08-04 08:33:54 +08:00
7eb27372b3
fix(server): capture legal multi-request prefill CUDA graph batches (#30206 )
Po-Han Huang (NVIDIA)
2026-08-04 07:56:59 +08:00
b819d2fb5b
[Docs] Rename docs_new/ to docs/ (#32123 )
zijiexia and Claude Opus 4.8
2026-08-03 16:51:00 -07:00
c949e91f18
[CI] Remove the orphaned site-packages sglang skeleton that shadows the checkout (#33441 )
Liangsheng Yin
2026-08-03 16:31:25 -07:00
22c2e2bcad
[CI] Persist the cargo build cache across CUDA CI jobs (#33361 )
Liangsheng Yin
2026-08-03 16:23:09 -07:00
b8f6181bff
[CI] Build the Rust extensions with the pinned toolchain instead of the image default (#33437 )
Liangsheng Yin
2026-08-03 16:16:34 -07:00
bc7e1a07c3
Bound prefill delayer all-branch delay and decay the max_prefill_bs high-watermark (#32880 )
Hanming Lu
2026-08-03 14:43:02 -07:00
7cd79fda56
[CI] Skip absent inline suites when loading timeouts (#33410 )
Xinyuan Tong
2026-08-04 05:21:25 +08:00
4ef1660cd8
[Docs] MiniMax-H3: add measured H200 Ulysses4 vs TP2+Ulysses2 topology data (#33398 )
Xiaoyu Zhang and Claude Fable 5
2026-08-04 03:54:29 +08:00
db0fe370b7
[PD] Fix false health-503 during decode retraction re-admission (#33118 )
cctry
2026-08-03 09:44:25 -07:00
6cf661117d
[PD] Add a queues.prealloc_ready counter to the load snapshot (#33133 )
cctry
2026-08-03 09:44:11 -07:00
0ba46c88e5
[diffusion] CI: add minimax-h3 2-gpu consistency coverage (#33281 )
Mick
2026-08-03 22:40:08 +08:00
3953788596
[HiSparse]Fix DeepSeek V4 HiSparse PD Transfers with Separate Host and Device KV Indices (#31901 )
huangtingwei and jackyYang6
2026-08-03 20:35:04 +08:00
d48ab2d386
Fix BCG circular import during server startup (#33371 )
Mohammad Miadh Angkad
2026-08-03 16:53:16 +08:00
a2d1003b18
[Mooncake] Fix ProcessGroup API imports (#32403 )
Xun Sun
2026-08-03 16:16:41 +08:00
92999d84f4
[AMD]Qwen3.5 integration gfx950 fmha fp8 hd256 (#32046 )
amd-oshkarav
2026-08-03 09:25:31 +02:00
45c00daa1b
[misc] Deep-merge nested config overrides and parse request bodies with orjson (#33351 )
Liangsheng Yin
2026-08-02 23:55:06 -07:00
204e0fbac0
[SM120] Only split touched SWA pages in FlashMLA page-split kernel (#32320 )
2026-08-03 14:45:07 +08:00
21d930aae3
[AMD] Fix JIT compile failure in sgl_kernel/warp.cuh (#33195 )
Wang, FangYuan and Bingxu Chen
2026-08-03 14:35:29 +08:00
e9366d7f79
Updated b200 kimi-k3 cookbook (#33038 )
2026-08-03 14:16:39 +08:00
e00f32ed4f
feat: rust sglang server openai apis (#33103 )
Chengyu Lin and Rain Jiang
2026-08-03 02:13:44 -04:00
0bf0640b9d
[CP] Support breakable CUDA graphs for zigzag strategy (#33136 )
Baizhou Zhang
2026-08-02 23:11:57 -07:00
85484c457d
docs: refresh README news highlights (#33347 )
Lianmin Zheng
2026-08-02 22:37:01 -07:00
fcc4de9e5f
[CI] Size the CPU stage from the live partition model (#33329 )
Liangsheng Yin
2026-08-02 22:26:58 -07:00
c2dbdfd218
[Docs] Fix overlapping quality-profile table headers on MiniMax-H3 page (#33345 )
zijiexia and Claude Opus 5
2026-08-02 22:22:03 -07:00
5fe97637df
Support DeepGEMM for standard MoE dispatch (#33128 )
YAMY and Sam Li
2026-08-02 21:48:13 -07:00
f5f021672a
[Fix] Treat an empty grammar constraint as unset in SamplingParams (#33328 )
Liangsheng Yin
2026-08-02 21:45:51 -07:00
b64fd800d4
docs(diffusion): update skills for MiniMax-H3 (#33282 )
Xiaoyu Zhang
2026-08-03 12:44:01 +08:00
b1754a8f3b
[gdn] fused replayssm ring write into flashinfer gdn mtp verify kernel (#33102 )
Qiaolin Yu
2026-08-02 21:31:22 -07:00
b8109b5d63
config: retire the last process-global config field reads (#33338 )
Cheng Wan
2026-08-02 21:24:42 -07:00
aa3bbbc6e8
observability: publish the generated forward-pass-metrics endpoint to the bags (#33337 )
Cheng Wan
2026-08-02 21:24:11 -07:00
0b3e8bedd1
config: keep runtime hicache and weight-version updates off ServerArgs (#33336 )
Cheng Wan
2026-08-02 21:23:38 -07:00
9bc8848fcf
spec: build every draft worker from a draft ServerArgs copy (#33335 )
Cheng Wan
2026-08-02 21:22:52 -07:00
ebb1c88d23
config: stop writing config onto the published ServerArgs at three sites (#33334 )
Cheng Wan
2026-08-02 21:22:05 -07:00
8186eeb939
[rust-server] Reland: fix TCP-layer TTFT stalls (#33026 ) (#33269 )
Kan Wu and Claude Fable 5
2026-08-02 20:52:29 -07:00
e824b24250
[CP] Fuse zigzag attention into a single call (#33137 )
Baizhou Zhang
2026-08-02 20:46:42 -07:00
741e33db81
fix(sampling): reject conflicting structural tag constraints (#32525 )
Mark
2026-08-03 11:24:55 +08:00
dd6ddc053b
[Fix] Drop deprecated multimodal processor residency state (#33308 )
Liangsheng Yin and Mick
2026-08-02 20:02:53 -07:00
28a2472f95
[DeepSeek-V4] Fix nvcc 13 crash building the topk_v2 kernel (#32910 )
2026-08-02 19:59:42 -07:00
4bc593fdc8
[Perf] Broadcast single-image DP vision embedding instead of pad-to-max all-gather (#33307 )
Liangsheng Yin and Mick
2026-08-02 19:39:53 -07:00
2a7a299c27
[Spec] Support sampling in the DSPARK graph-folded draft proposal (#33298 )
Liangsheng Yin
2026-08-02 18:46:39 -07:00
5d2dbb35a6
[rust-server] PD disaggregation support (#33125 )
2026-08-02 18:32:59 -07:00
c844244da5
support dp attn with client lb (#33105 )
Rain Jiang and Kan Wu
2026-08-02 18:32:35 -07:00
12eadf86f1
[AMD] Enable mamba JIT transfer kernel on ROCm (fix transfer_kv_mamba NameError) (#31741 )
kangwangamd
2026-08-03 07:40:17 +08:00
88e5a0f635
test: stand up the config tiers two unit tests read from (#33294 )
Cheng Wan
2026-08-02 15:46:34 -07:00
8cc941a672
[BCG][4/N] Enable bcg on megamoe & flashinfer a2a backend (#33150 )
Yuwei An and Claude Opus 5
2026-08-02 14:56:03 -07:00
ec741e4161
Fix DSpark loading for hybrid DSV4 NVFP4 (#33276 )
Mohammad Miadh Angkad
2026-08-03 02:05:09 +08:00
1a3bea77f2
[Feat] DCP + HiCache L2 Support (ported from kimi-k3) (#33112 )
Yuwei An and Claude Opus 5
2026-08-02 08:35:27 -07:00
f8e62a9224
[NPU] Add PR test cases (#32392 )
2026-08-02 23:00:02 +08:00
27b15349e5
[CI] Fix stale CPU test fixtures (#33277 )
Mohammad Miadh Angkad
2026-08-02 22:38:45 +08:00
70fe2e0dd5
[diffusion] model: support minimax-h3 (#33275 )
2026-08-02 22:32:37 +08:00
0877a0e2f1
[CI] Add speculative_draft_attention_backend to the page-constraint test view (#33254 )
Alison Shao
2026-08-02 04:58:59 -07:00
558c9bdcc2
[misc] Improve benchmark determinism and dataset API coverage (#33255 )
Liangsheng Yin
2026-08-02 01:39:50 -07:00
06554515f4
Revert "[rust-server] fix TCP-layer TTFT stalls" (#33257 )
Kan Wu
2026-08-02 01:20:11 -07:00
43be25b2b7
[CI] Graceful teardown for kv_canary and EAGLE spec fixtures (#32829 )
Alison Shao
2026-08-02 01:09:59 -07:00
8d106c3d79
[NPU] Enable automatic ascend_attn selection for vision attention and graph runners (#31948 )
2026-08-02 15:09:51 +08:00
a0b7bcf592
[Feature] Support return_hidden_states="last" (#30177 )
Tao Li and litao.dream
2026-08-02 15:09:33 +08:00
1685d29f21
[AMD] Fix DeepSeek-V4 fused-RMS FP8 scale metadata on gfx950 (#31727 )
sonle5
2026-08-02 14:06:09 +07:00
21d932069b
docs: drop unreachable inkling LoRA benchmark entry (#33251 )
zijiexia and Claude Opus 5
2026-08-01 23:58:35 -07:00
37be4e9247
[AMD] Speed up DSV4 MoE weight loading from mmap views (#32315 )
Bingxu Chen
2026-08-02 14:47:47 +08:00
7e509f690e
[AMD] Fix DeepSeek-V4 FP4 MoE expert memory bloat (#31450 )
Thomas Wang
2026-08-02 14:42:21 +08:00
056474cdb0
feat: Session-reference-aware Unified Radix Cache for agentic multi-turn workloads (#29173 )
2026-08-02 13:43:55 +08:00
131bd51b01
[Spec] Add trtllm_mha support for Gemma 4 MTP draft attention backend (#25545 )
2026-08-01 16:16:04 -07:00
e2cf21b9e5
[Kimi K3] Add reasoning, tool-call, and OpenAI serving support (#33025 )
2026-08-02 05:57:23 +08:00
f1b41a5b3d
[CP]: FIx some issue for glm5.2 cp v2 (#33100 )
Zhangheng
2026-08-02 04:50:10 +08:00
e0ba311026
Disable breakable CUDA graph for NemotronH (#33130 )
Brayden Zhong
2026-08-01 13:48:58 -07:00
574ead753a
[rust-server] fix TCP-layer TTFT stalls (#33026 )
2026-08-01 12:13:13 -07:00
00a219f6c9
[Quant] Keep the flashinfer_deepgemm FP8 GEMM to 1 <= M < 32 (#32843 )
Jackey Hua and Claude Opus 5
2026-08-01 09:36:13 -07:00
729081e14f
docs: rewrite the runtime-context skill for the namespace-bag config model (#33173 )
Cheng Wan
2026-08-01 09:00:10 -07:00
7071cfb873
runtime_context: per-role namespace enforcement behind SGLANG_ROLE_NAMESPACES (#33172 )
Cheng Wan
2026-08-01 08:59:49 -07:00