diff --git a/server/text_generation_server/models/flash_causal_lm.py b/server/text_generation_server/models/flash_causal_lm.py
index e68a2100..21ed4f6c 100644
--- a/server/text_generation_server/models/flash_causal_lm.py
+++ b/server/text_generation_server/models/flash_causal_lm.py
@@ -777,7 +777,7 @@ class FlashCausalLM(Model):
             self.device,
         )
 
-        if os.getenv("ENABLE_CUDA_GRAPHS", "false") == "true":
+        if os.getenv("ENABLE_CUDA_GRAPHS", "False") == "True":
             try:
                 # Warmup cuda graphs for all power of twos until 64
                 for i in range(6):