diff --git a/src/main/utils/llamacpp.ts b/src/main/utils/llamacpp.ts
index 051d3ac95..f2881fa8b 100644
--- a/src/main/utils/llamacpp.ts
+++ b/src/main/utils/llamacpp.ts
@@ -165,6 +165,8 @@ const getAssetPattern = (tag: string, variant: string): { pattern: string; isZip
       cpu: `llama-${tag}-bin-win-cpu-${archStr}.zip`,
       'cuda-12.4': `llama-${tag}-bin-win-cuda-12.4-x64.zip`,
       'cuda-13.1': `llama-${tag}-bin-win-cuda-13.1-x64.zip`,
+      'cuda-13.3': `llama-${tag}-bin-win-cuda-13.3-x64.zip`,   // new
+      'cuda-13.4': `llama-${tag}-bin-win-cuda-13.4-x64.zip`,   // new
       vulkan: `llama-${tag}-bin-win-vulkan-x64.zip`
     }
     const name = variantMap[variant] ?? variantMap.cpu
diff --git a/src/renderer/src/lib/components/Main/Settings/InferenceRuntime.svelte b/src/renderer/src/lib/components/Main/Settings/InferenceRuntime.svelte
index a683160eb..b0b684e6f 100644
--- a/src/renderer/src/lib/components/Main/Settings/InferenceRuntime.svelte
+++ b/src/renderer/src/lib/components/Main/Settings/InferenceRuntime.svelte
@@ -61,6 +61,8 @@
       { value: 'cpu', label: $i18n.t('settings.inference.variantCPU') },
       { value: 'cuda-12.4', label: 'CUDA 12.4' },
       { value: 'cuda-13.1', label: 'CUDA 13.1' },
+      { value: 'cuda-13.3', label: 'CUDA 13.3' },
+      { value: 'cuda-13.4', label: 'CUDA 13.4' },
       { value: 'vulkan', label: 'Vulkan' }
     ]
     return [