Gate supports_nvfp4_compute on CUDA 13+ build (fixes #11864)

cuBLAS FP4 matmul kernels (cublasLtMatmulAlgoGetHeuristic) require
CUDA 13.0+. On Blackwell GPUs with torch built against CUDA <13
(e.g. cu128), the previous check only looked at compute capability
and let native NVFP4 through, causing CUBLAS_STATUS_NOT_SUPPORTED
errors or VRAM blowups at matmul time. Now falls back to the
regular quantized-storage path when the CUDA build is too old.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

comfy/model_management.py

@@ -1963,6 +1963,14 @@ def supports_nvfp4_compute(device=None):
     if props.major < 10:
         return False
 
+    # cuBLAS FP4 matmul kernels require CUDA 13+, see #11864
+    try:
+        cuda_version_major = int(torch.version.cuda.split(".")[0])
+    except:
+        return False
+    if cuda_version_major < 13:
+        return False
+
     return True
 
 def supports_mxfp8_compute(device=None):