Commits · b0959cfdbce3dd11273449e8c4a89d3506640f09 · OpenDAS / dynamo

"components/vscode:/vscode.git/clone" did not exist on "79ad7f36dd4b8f96af8839146c1783035a305ea1"

21 Jan, 2026 2 commits
- feat: warmup dataset for planner load predictor (#5529) · b0959cfd
  Hongkuan Zhou authored Jan 21, 2026
```
Signed-off-by: hongkuanz <hongkuanz@nvidia.com>
```
  b0959cfd
- fix: multi image bug for qwen ec connector (#5514) · 60700899
  Ayush Agarwal authored Jan 21, 2026
```
Signed-off-by: ayushag <ayushag@nvidia.com>
```
  60700899
20 Jan, 2026 7 commits
- fix: ipv6 support for sglang disaggregation (#5521) · 9b8b35a8
  ishandhanani authored Jan 20, 2026
```
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
```
  9b8b35a8
- refactor: clean up SGLang sleep/wake implementation (#5517) · 2e8c4447
  Schwinn Saereesitthipitak authored Jan 20, 2026
  
  2e8c4447
- fix(sglang): Fix YAML config parsing for store_true arguments (#5513) · 04cecda7
  ishandhanani authored Jan 20, 2026
  
  04cecda7
- feat(sglang): enforce stream_output=True for optimal streaming performance (#5510) · 748fee6b
  MatejKosec authored Jan 20, 2026
```
This ensures that only new tokens are returned by sglang which avoids the overhead from creating copies of the entire token sequences per each iteration. These copies can become a bottleneck particularly for long sequence lengths and large concurrency counts.
Signed-off-by: Matej Kosec <mkosec@nvidia.com>
```
  748fee6b
- feat: Support Dynamo KVBM with TRTLLM Disagg (#3527) · bf19823d
  jthomson04 authored Jan 20, 2026
```
Signed-off-by: jthomson04 <jwillthomson19@gmail.com>
```
  bf19823d
- feat: SGLang release/resume_memory_occupation endpoints (#5207) · 0e0d6c16
  Schwinn Saereesitthipitak authored Jan 20, 2026
```
Signed-off-by: Schwinn Saereesitthipitak <schwinns@nvidia.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
```
  0e0d6c16
- feat: Sleep/wake endpoints for vLLM runtime (#5339) · e2a0a4b5
  Schwinn Saereesitthipitak authored Jan 20, 2026
```
Signed-off-by: Schwinn Saereesitthipitak <schwinns@nvidia.com>
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
```
  e2a0a4b5
18 Jan, 2026 1 commit
- feat: track output tokens / blocks in the Router (optional) (#5452) · 050906b5
  Yan Ru Pei authored Jan 18, 2026
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  050906b5
16 Jan, 2026 2 commits
- chore: support for muliple MM requests for vLLM Encoder + Rust Based Preprocessor (#5463) · f9918f61
  Ayush Agarwal authored Jan 16, 2026
```
Signed-off-by: ayushag <ayushag@nvidia.com>
```
  f9918f61
- feat: SGLang aggregated multimodal support (#5450) · ddee21cb
  KrishnanPrash authored Jan 16, 2026
  
  ddee21cb
15 Jan, 2026 3 commits
- fix: [vLLM multimodel] launch image loading in parallel (#5444) · 0b7e1271
  GuanLuo authored Jan 15, 2026
```
Signed-off-by: Guan Luo <41310872+GuanLuo@users.noreply.github.com>
```
  0b7e1271
- fix: consolidate return statement in get_engine_runtime_config (#5442) · d1545621
  dagil-nvidia authored Jan 15, 2026
```
Signed-off-by: Dan Gil <dagil@nvidia.com>
```
  d1545621
- fix: rename test file to skip vllm specific tests (#5438) · d02619b8
  jh-nv authored Jan 15, 2026
  
  d02619b8
14 Jan, 2026 3 commits
- chore(router): Remove deprecated best_worker_id and metrics_labels (#5412) · c5458b99
  Graham King authored Jan 14, 2026
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  c5458b99
- fix: Properly handle multiple text components from request (#5196) · e22bb037
  jh-nv authored Jan 14, 2026
  
  e22bb037
- refactor: vllm EPD refactor (#4994) · 334cbd9b
  GuanLuo authored Jan 14, 2026
```
Signed-off-by: Guan Luo <gluo@nvidia.com>
Signed-off-by: Guan Luo <41310872+GuanLuo@users.noreply.github.com>
```
  334cbd9b
13 Jan, 2026 1 commit
- feat: Propagate OTEL tracing context for trtllm (#5377) · 8c129ed4
  jh-nv authored Jan 13, 2026
  
  8c129ed4
12 Jan, 2026 3 commits
- feat: allow router not assuming decode kv reuse (#5350) · 7fdc742e
  Yan Ru Pei authored Jan 12, 2026
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  7fdc742e
- fix: pass back `usage` when using text mode with VLLM (#5336) · 68235126
  Qidong Su authored Jan 12, 2026
```
Signed-off-by: Qidong Su <soodoshll@gmail.com>
```
  68235126
- fix(sglang): use external_trace_header API for distributed tracing (#5346) · cbbaa6b7
  ishandhanani authored Jan 13, 2026
```
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
```
  cbbaa6b7
09 Jan, 2026 2 commits
- fix: Fix decode worker in vllm for qwen_vl models (#5281) · 5cd8005c
  Indrajit Bhosale authored Jan 09, 2026
```
Signed-off-by: Krishnan Prashanth <kprashanth@nvidia.com>
Co-authored-by: Krishnan Prashanth <kprashanth@nvidia.com>
```
  5cd8005c
- fix: enable toggling kv events pub/sub (currently nats based) with --no-kv-events flag (#5237) · 0d3ff440
  Biswa Panda authored Jan 08, 2026
  
  0d3ff440
08 Jan, 2026 3 commits
- chore: move custom mm preprocessor to new file (#5278) · 1c9fea01
  Ayush Agarwal authored Jan 08, 2026
```
Signed-off-by: ayushag <ayushag@nvidia.com>
```
  1c9fea01
- feat(vllm): Add prompt embeds support for pre-computed inference inputs (#4739) · 6f9619a2
  Wenqi Glantz authored Jan 08, 2026
```
Signed-off-by: Wenqi Glantz <wglantz@nvidia.com>
```
  6f9619a2
- fix(vllmsglang): Give sglang/vllm the HF model name not the full path (#5274) · fcc4a60f
  Graham King authored Jan 08, 2026
```
Signed-off-by: Graham King <grahamk@nvidia.com>
```
  fcc4a60f
07 Jan, 2026 4 commits
- revert: remove vLLM backend metrics from planner (#4134) (#5185) · daa1d740
  daiyaanarfeen authored Jan 07, 2026
```
Signed-off-by: Daiyaan <darfeen@nvidia.com>
```
  daa1d740
- feat: ec connector handler (#5162) · 85e0512f
  Ayush Agarwal authored Jan 07, 2026
```
Signed-off-by: Ayush Agarwal <ayushag@nvidia.com>
```
  85e0512f
- fix: Optimize TRTLLM multimodal request processing by reusing the tokenizer (#5217) · 535528a5
  Indrajit Bhosale authored Jan 07, 2026
  
  535528a5
- fix: KvCacheConfig Settings Lost When Publishing Events (#5198) · 66dfc494
  Indrajit Bhosale authored Jan 07, 2026
  
  66dfc494
06 Jan, 2026 2 commits
- fix: update vLLM to 0.13.0 with API compatibility fixes (#5222) · a1333a8d
  Alec authored Jan 06, 2026
```
Signed-off-by: Vasilis Vagias <vvagias@nvidia.com>
Signed-off-by: alec-flowers <aflowers@nvidia.com>
Co-authored-by: Vasilis Vagias <vvagias@nvidia.com>
```
  a1333a8d
- fix: handle concurrent load lora calls (#5184) · 07721d1c
  Biswa Panda authored Jan 05, 2026
  
  07721d1c
05 Jan, 2026 1 commit
- fix: the error when encountering pure text conversations (#5088) · ea950355
  shpgy-shpgy authored Jan 06, 2026
```
Signed-off-by: shpgy-shpgy <875664365@qq.com>
```
  ea950355
04 Jan, 2026 1 commit
- fix: sglang metrics and prefill router fix (#5147) · 007c5b60
  ishandhanani authored Jan 04, 2026
  
  007c5b60
03 Jan, 2026 2 commits
- fix(sglang): expose TokenizerMetricsCollector metrics via Prometheus (#5120) · 3e341fd6
  ishandhanani authored Jan 03, 2026
  
  3e341fd6
- feat: mockers with bootstrap optimization (sglang testing) + CI test (#5121) · 0980b27f
  Yan Ru Pei authored Jan 02, 2026
```
Signed-off-by: PeaBrane <yanrpei@gmail.com>
```
  0980b27f
02 Jan, 2026 1 commit
- chore: update all copyright headers in repo to 2026 (#5130) · cf433e68
  Tushar Sharma authored Jan 02, 2026
```
Signed-off-by: Tushar Sharma <tusharma@nvidia.com>
```
  cf433e68
31 Dec, 2025 1 commit

fix: sglang disagg routing fixes and optimizations [DYN-1692] (#5106) · 0b33c1df

Yan Ru Pei authored Dec 31, 2025


Signed-off-by: PeaBrane <yanrpei@gmail.com>
Co-authored-by: Ishan Dhanani <ishandhanani@gmail.com>
Co-authored-by: Sean SH Choi <sechoi@nvidia.com>
Co-authored-by: ishandhanani <82981111+ishandhanani@users.noreply.github.com>

0b33c1df

30 Dec, 2025 1 commit

fix: Fix TRTLLM disagg with kv router (#4950) · d7c27e68

jthomson04 authored Dec 29, 2025


Signed-off-by: jthomson04 <jwillthomson19@gmail.com>
Co-authored-by: Ryan McCormick <rmccormick@nvidia.com>

d7c27e68