diff --git a/.github/workflows/e2e-aws-custom.yml b/.github/workflows/e2e-aws-custom.yml
index 970ffb78f2..7960732902 100644
--- a/.github/workflows/e2e-aws-custom.yml
+++ b/.github/workflows/e2e-aws-custom.yml
@@ -148,7 +148,7 @@ jobs:
           nvidia-smi
           python3.11 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed
diff --git a/.github/workflows/e2e-nvidia-l4-x1.yml b/.github/workflows/e2e-nvidia-l4-x1.yml
index a4f59abeab..9b2063b3d3 100644
--- a/.github/workflows/e2e-nvidia-l4-x1.yml
+++ b/.github/workflows/e2e-nvidia-l4-x1.yml
@@ -129,7 +129,7 @@ jobs:
           nvidia-smi
           python3.11 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed
diff --git a/.github/workflows/e2e-nvidia-l40s-x4-llama.yml b/.github/workflows/e2e-nvidia-l40s-x4-llama.yml
index a72ab8049a..cbb075e202 100644
--- a/.github/workflows/e2e-nvidia-l40s-x4-llama.yml
+++ b/.github/workflows/e2e-nvidia-l40s-x4-llama.yml
@@ -171,7 +171,7 @@ jobs:
           nvidia-smi
           python3.11 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed
diff --git a/.github/workflows/e2e-nvidia-l40s-x4-py312.yml b/.github/workflows/e2e-nvidia-l40s-x4-py312.yml
index d0fce0c663..705a802e52 100644
--- a/.github/workflows/e2e-nvidia-l40s-x4-py312.yml
+++ b/.github/workflows/e2e-nvidia-l40s-x4-py312.yml
@@ -141,7 +141,7 @@ jobs:
           nvidia-smi
           python3.12 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.12 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.12 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed
diff --git a/.github/workflows/e2e-nvidia-l40s-x4.yml b/.github/workflows/e2e-nvidia-l40s-x4.yml
index d481820e9d..6c7f3c3dd7 100644
--- a/.github/workflows/e2e-nvidia-l40s-x4.yml
+++ b/.github/workflows/e2e-nvidia-l40s-x4.yml
@@ -141,7 +141,7 @@ jobs:
           nvidia-smi
           python3.11 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed
diff --git a/.github/workflows/e2e-nvidia-l40s-x8.yml b/.github/workflows/e2e-nvidia-l40s-x8.yml
index 5b69c1282e..a4308f3f3d 100644
--- a/.github/workflows/e2e-nvidia-l40s-x8.yml
+++ b/.github/workflows/e2e-nvidia-l40s-x8.yml
@@ -221,7 +221,7 @@ jobs:
           nvidia-smi
           python3.11 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed
diff --git a/.github/workflows/e2e-nvidia-t4-x1.yml b/.github/workflows/e2e-nvidia-t4-x1.yml
index c3325f70c0..07b4a9a59f 100644
--- a/.github/workflows/e2e-nvidia-t4-x1.yml
+++ b/.github/workflows/e2e-nvidia-t4-x1.yml
@@ -136,7 +136,7 @@ jobs:
           nvidia-smi
           python3.11 -m pip cache remove llama_cpp_python
 
-          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v .
+          CMAKE_ARGS="-DGGML_CUDA=on" python3.11 -m pip install -v . -c constraints-dev.txt
 
           # https://github.com/instructlab/instructlab/issues/1821
           # install with Torch and build dependencies installed