diff --git a/local-inference/requirements.in b/local-inference/requirements.in
index 617073dfc..db90c08dc 100644
--- a/local-inference/requirements.in
+++ b/local-inference/requirements.in
@@ -1,5 +1,5 @@
 accelerate==0.33.0
 torch==2.4.0
 transformers==4.43.3
-quart==0.19.6
+quart==0.19.7
 quart_cors==0.7.0
\ No newline at end of file
diff --git a/local-inference/requirements.txt b/local-inference/requirements.txt
index 7d0680686..b02d47d9b 100644
--- a/local-inference/requirements.txt
+++ b/local-inference/requirements.txt
@@ -21,6 +21,7 @@ filelock==3.15.4
     #   huggingface-hub
     #   torch
     #   transformers
+    #   triton
 flask==3.0.3
     # via quart
 fsspec==2024.6.1
@@ -68,6 +69,37 @@ numpy==1.26.4
     # via
     #   accelerate
     #   transformers
+nvidia-cublas-cu12==12.1.3.1
+    # via
+    #   nvidia-cudnn-cu12
+    #   nvidia-cusolver-cu12
+    #   torch
+nvidia-cuda-cupti-cu12==12.1.105
+    # via torch
+nvidia-cuda-nvrtc-cu12==12.1.105
+    # via torch
+nvidia-cuda-runtime-cu12==12.1.105
+    # via torch
+nvidia-cudnn-cu12==9.1.0.70
+    # via torch
+nvidia-cufft-cu12==11.0.2.54
+    # via torch
+nvidia-curand-cu12==10.3.2.106
+    # via torch
+nvidia-cusolver-cu12==11.4.5.107
+    # via torch
+nvidia-cusparse-cu12==12.1.0.106
+    # via
+    #   nvidia-cusolver-cu12
+    #   torch
+nvidia-nccl-cu12==2.20.5
+    # via torch
+nvidia-nvjitlink-cu12==12.6.77
+    # via
+    #   nvidia-cusolver-cu12
+    #   nvidia-cusparse-cu12
+nvidia-nvtx-cu12==12.1.105
+    # via torch
 packaging==24.1
     # via
     #   accelerate
@@ -82,7 +114,7 @@ pyyaml==6.0.1
     #   accelerate
     #   huggingface-hub
     #   transformers
-quart==0.19.6
+quart==0.19.7
     # via
     #   -r requirements.in
     #   quart-cors
@@ -98,8 +130,6 @@ safetensors==0.4.3
     # via
     #   accelerate
     #   transformers
-setuptools==72.1.0
-    # via torch
 sympy==1.13.1
     # via torch
 tokenizers==0.19.1
@@ -114,6 +144,8 @@ tqdm==4.66.4
     #   transformers
 transformers==4.43.3
     # via -r requirements.in
+triton==3.0.0
+    # via torch
 typing-extensions==4.12.2
     # via
     #   huggingface-hub