diff --git a/discover/path.go b/discover/path.go
index b6b648dbf..23aa8110d 100644
--- a/discover/path.go
+++ b/discover/path.go
@@ -19,11 +19,6 @@ var LibOllamaPath string = func() string {
 		return ""
 	}
 
-	exe, err = filepath.EvalSymlinks(exe)
-	if err != nil {
-		return ""
-	}
-
 	var libPath string
 	switch runtime.GOOS {
 	case "windows":
diff --git a/llm/server.go b/llm/server.go
index 50ba91f18..f88963060 100644
--- a/llm/server.go
+++ b/llm/server.go
@@ -320,11 +320,6 @@ func NewLlamaServer(gpus discover.GpuInfoList, model string, f *ggml.GGML, adapt
 			return nil, fmt.Errorf("unable to lookup executable path: %w", err)
 		}
 
-		exe, err = filepath.EvalSymlinks(exe)
-		if err != nil {
-			return nil, fmt.Errorf("unable to evaluate symlinks for executable path: %w", err)
-		}
-
 		// TODO - once fully switched to the Go runner, load the model here for tokenize/detokenize cgo access
 		s := &llmServer{
 			port:        port,