diff --git a/examples/imagenet/requirements.txt b/examples/imagenet/requirements.txt
index c2fd33dbd..3c714deeb 100644
--- a/examples/imagenet/requirements.txt
+++ b/examples/imagenet/requirements.txt
@@ -7,5 +7,5 @@ jax[cuda11_cudnn805]>=0.3.16  # change to jax[tpu] if running on tpus
 ml-collections==0.1.0
 numpy==1.22.0
 optax==0.1.3
-tensorflow==2.11.1
+tensorflow==2.12.1
 tensorflow-datasets==4.4.0
diff --git a/examples/lm1b/requirements.txt b/examples/lm1b/requirements.txt
index 8ebc88977..599cbb567 100644
--- a/examples/lm1b/requirements.txt
+++ b/examples/lm1b/requirements.txt
@@ -7,7 +7,7 @@ jaxlib==0.4.13+cuda11.cudnn82  # Make sure CUDA version matches the base image.
 ml-collections==0.1.1
 numpy==1.24.3
 optax==0.1.5
-sentencepiece==0.1.99
-tensorflow==2.13.0
+sentencepiece==0.2.1
+tensorflow==2.21.0
 tensorflow-datasets==4.9.2
 tensorflow-text==2.13.0
diff --git a/examples/mnist/requirements.txt b/examples/mnist/requirements.txt
index c622dd821..7712c34f9 100644
--- a/examples/mnist/requirements.txt
+++ b/examples/mnist/requirements.txt
@@ -7,5 +7,5 @@ jaxlib==0.3.2+cuda11.cudnn82  # Make sure CUDA version matches the base image.
 ml-collections==0.1.0
 numpy==1.22.0
 optax==0.1.0
-tensorflow==2.11.1
+tensorflow==2.12.1
 tensorflow-datasets==4.4.0
diff --git a/examples/nlp_seq/requirements.txt b/examples/nlp_seq/requirements.txt
index d7c43cb0c..c715ddd16 100644
--- a/examples/nlp_seq/requirements.txt
+++ b/examples/nlp_seq/requirements.txt
@@ -1,4 +1,4 @@
 absl-py==1.0.0
 flax==0.3.6
 numpy==1.22.0
-tensorflow==2.11.1
+tensorflow==2.12.1
diff --git a/examples/ogbg_molpcba/requirements.txt b/examples/ogbg_molpcba/requirements.txt
index 8707b9e0d..52e1bfad0 100644
--- a/examples/ogbg_molpcba/requirements.txt
+++ b/examples/ogbg_molpcba/requirements.txt
@@ -9,5 +9,5 @@ ml-collections==0.1.0
 numpy==1.22.0
 optax==0.1.0
 sklearn==0.0
-tensorflow==2.11.1
+tensorflow==2.12.1
 tensorflow-datasets==4.4.0
diff --git a/examples/ppo/requirements.txt b/examples/ppo/requirements.txt
index c1dd9c894..f2dfad9ba 100644
--- a/examples/ppo/requirements.txt
+++ b/examples/ppo/requirements.txt
@@ -1,10 +1,10 @@
 absl-py==1.0.0
 atari-py==0.2.5
-opencv-python==4.5.4.60
+opencv-python==4.8.1.78
 flax==0.3.6
 gym==0.18.3
 gymnasium[atari, accept-rom-license]==0.29.0
 ml-collections==0.1.0
 numpy==1.22.0
 optax==0.1.5
-tensorflow==2.11.1
+tensorflow==2.12.1
diff --git a/examples/sst2/requirements.txt b/examples/sst2/requirements.txt
index e82176489..4e329cbb8 100644
--- a/examples/sst2/requirements.txt
+++ b/examples/sst2/requirements.txt
@@ -4,6 +4,6 @@ flax==0.3.6
 ml-collections==0.1.0
 numpy==1.22.0
 optax==0.1.0
-tensorflow==2.11.1
+tensorflow==2.12.1
 tensorflow-datasets==4.4.0
 tensorflow-text==2.7.0
diff --git a/examples/vae/requirements.txt b/examples/vae/requirements.txt
index cc497ca7e..2b11ff259 100644
--- a/examples/vae/requirements.txt
+++ b/examples/vae/requirements.txt
@@ -2,6 +2,6 @@ absl-py==1.4.0
 flax==0.6.9
 numpy==1.23.5
 optax==0.1.5
-Pillow==10.2.0
-tensorflow==2.12.0
+Pillow==12.2.0
+tensorflow==2.12.1
 tensorflow-datasets==4.9.2
\ No newline at end of file
diff --git a/examples/wmt/requirements.txt b/examples/wmt/requirements.txt
index 616db2d1b..f82cfa729 100644
--- a/examples/wmt/requirements.txt
+++ b/examples/wmt/requirements.txt
@@ -7,8 +7,8 @@ jax[cuda11_cudnn805]>=0.3.16  # change to jax[tpu] if running on tpus
 ml-collections==0.1.0
 numpy==1.22.0
 optax==0.1.0
-sentencepiece==0.1.96
+sentencepiece==0.2.1
 six==1.15.0
-tensorflow==2.11.1
+tensorflow==2.12.1
 tensorflow-datasets==4.4.0
 tensorflow-text==2.8.1
diff --git a/pyproject.toml b/pyproject.toml
index aefae59d4..ff7275e4b 100644
--- a/pyproject.toml
+++ b/pyproject.toml
@@ -56,7 +56,7 @@ testing = [
     "pytest-xdist",
     "pytype",
     # WMT/LM1B examples
-    "sentencepiece==0.2.0",  # Segfault bug in 0.2.1
+    "sentencepiece==0.2.1",  # Segfault bug in 0.2.1
     "tensorflow_text>=2.11.0; platform_system!='Darwin' and python_version < '3.13'",
     "tensorflow_datasets",
     "tensorflow_cpu>=2.12.0; python_version<'3.13'", # to fix Numpy np.bool8 deprecation error