chore: bump SGLang version to 0.5.6 (#14316)

Co-authored-by: sglang-bot <sglang-bot@users.noreply.github.com>

benchmark/deepseek_v3/README.md

@@ -33,7 +33,7 @@ Add [performance optimization options](#performance-optimization-options) as nee
 
 ```bash
 # Installation
-pip install "sglang[all]>=0.5.5.post3"
+pip install "sglang[all]>=0.5.6"
 
 # Launch
 python3 -m sglang.launch_server --model deepseek-ai/DeepSeek-V3 --tp 8 --trust-remote-code

docker/Dockerfile

@@ -10,7 +10,7 @@ ARG GRACE_BLACKWELL_DEEPEP_BRANCH=gb200_blog_part_2
 ARG DEEPEP_COMMIT=9af0e0d0e74f3577af1979c9b9e1ac2cad0104ee
 ARG BUILD_AND_DOWNLOAD_PARALLEL=8
 ARG SGL_KERNEL_VERSION=0.3.18.post2
-ARG SGL_VERSION=0.5.5.post3
+ARG SGL_VERSION=0.5.6
 ARG USE_LATEST_SGLANG=0
 ARG GDRCOPY_VERSION=2.5.1
 ARG PIP_DEFAULT_INDEX

docker/rocm.Dockerfile

@@ -1,7 +1,7 @@
 # Usage (to build SGLang ROCm docker image):
-#   docker build --build-arg SGL_BRANCH=v0.5.5.post3 --build-arg GPU_ARCH=gfx942 -t v0.5.5.post3-rocm630-mi30x -f rocm.Dockerfile .
+#   docker build --build-arg SGL_BRANCH=v0.5.6 --build-arg GPU_ARCH=gfx942 -t v0.5.6-rocm630-mi30x -f rocm.Dockerfile .
-#   docker build --build-arg SGL_BRANCH=v0.5.5.post3 --build-arg GPU_ARCH=gfx942-rocm700 -t v0.5.5.post3-rocm700-mi30x -f rocm.Dockerfile .
+#   docker build --build-arg SGL_BRANCH=v0.5.6 --build-arg GPU_ARCH=gfx942-rocm700 -t v0.5.6-rocm700-mi30x -f rocm.Dockerfile .
-#   docker build --build-arg SGL_BRANCH=v0.5.5.post3 --build-arg GPU_ARCH=gfx950 -t v0.5.5.post3-rocm700-mi35x -f rocm.Dockerfile .
+#   docker build --build-arg SGL_BRANCH=v0.5.6 --build-arg GPU_ARCH=gfx950 -t v0.5.6-rocm700-mi35x -f rocm.Dockerfile .
 
 
 # Default base images

docs/get_started/install.md

@@ -25,7 +25,7 @@ uv pip install "sglang" --prerelease=allow
 
 ```bash
 # Use the last release branch
-git clone -b v0.5.5.post3 https://github.com/sgl-project/sglang.git
+git clone -b v0.5.6 https://github.com/sgl-project/sglang.git
 cd sglang
 
 # Install the python packages

docs/platforms/amd_gpu.md

@@ -44,7 +44,7 @@ You can install SGLang using one of the methods below.
 
 ```bash
 # Use the last release branch
-git clone -b v0.5.5.post3 https://github.com/sgl-project/sglang.git
+git clone -b v0.5.6 https://github.com/sgl-project/sglang.git
 cd sglang
 
 # Compile sgl-kernel

docs/platforms/ascend_npu.md

@@ -81,7 +81,7 @@ We are also providing a DeepEP-compatible Library as a drop-in replacement of de
 
 ```shell
 # Use the last release branch
-git clone -b v0.5.5.post3 https://github.com/sgl-project/sglang.git
+git clone -b v0.5.6 https://github.com/sgl-project/sglang.git
 cd sglang
 
 pip install --upgrade pip

python/pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sglang"
-version = "0.5.5.post3"
+version = "0.5.6"
 description = "SGLang is a fast serving framework for large language models and vision language models."
 readme = "README.md"
 requires-python = ">=3.10"

python/pyproject_cpu.toml

@@ -5,7 +5,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sglang"
-version = "0.5.5.post3"
+version = "0.5.6"
 description = "SGLang is a fast serving framework for large language models and vision language models."
 readme = "README.md"
 requires-python = ">=3.10"

python/pyproject_other.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sglang"
-version = "0.5.5.post3"
+version = "0.5.6"
 description = "SGLang is a fast serving framework for large language models and vision language models."
 readme = "README.md"
 requires-python = ">=3.10"

python/pyproject_xpu.toml

@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
 
 [project]
 name = "sglang"
-version = "0.5.5.post3"
+version = "0.5.6"
 description = "SGLang is a fast serving framework for large language models and vision language models."
 readme = "README.md"
 requires-python = ">=3.10"

python/sglang/version.py

@@ -1 +1 @@
-__version__ = "0.5.5.post3"
+__version__ = "0.5.6"

test/nightly/test_deepseek_v32_nsabackend.py

@@ -197,6 +197,7 @@ class TestDeepseekV32NasBackend_fp8kvcache(CustomTestCase):
         self.assertGreater(metrics["accuracy"], 0.935)
 
 
+@unittest.skip("Temporary skip pure TP test")
 class TestDeepseekV32NasBackend_pure_tp(CustomTestCase):
     """Test DeepSeek V3.2 with pure TP mode (no DP attention)."""