diff --git a/core/gallery/importers/importers_test.go b/core/gallery/importers/importers_test.go
index 48fb6ac3d46a..34814fe66c0d 100644
--- a/core/gallery/importers/importers_test.go
+++ b/core/gallery/importers/importers_test.go
@@ -27,6 +27,48 @@ var _ = Describe("DiscoverModelConfig", func() {
 			Expect(modelConfig.Files[0].URI).To(Equal("https://huggingface.co/mudler/LocalAI-functioncall-qwen2.5-7b-v0.5-Q4_K_M-GGUF/resolve/main/localai-functioncall-qwen2.5-7b-v0.5-q4_k_m.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
 			Expect(modelConfig.Files[0].SHA256).To(Equal("4e7b7fe1d54b881f1ef90799219dc6cc285d29db24f559c8998d1addb35713d4"), fmt.Sprintf("Model config: %+v", modelConfig))
 		})
+
+		It("should discover and import using LlamaCPPImporter", func() {
+			uri := "https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF"
+			preferences := json.RawMessage(`{}`)
+
+			modelConfig, err := importers.DiscoverModelConfig(uri, preferences)
+
+			Expect(err).ToNot(HaveOccurred(), fmt.Sprintf("Error: %v", err))
+			Expect(modelConfig.Name).To(Equal("Qwen3-VL-2B-Instruct-GGUF"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Description).To(Equal("Imported from https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.ConfigFile).To(ContainSubstring("backend: llama-cpp"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.ConfigFile).To(ContainSubstring("mmproj: mmproj/mmproj-Qwen3VL-2B-Instruct-Q8_0.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.ConfigFile).To(ContainSubstring("model: Qwen3VL-2B-Instruct-Q4_K_M.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(len(modelConfig.Files)).To(Equal(2), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[0].Filename).To(Equal("Qwen3VL-2B-Instruct-Q4_K_M.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[0].URI).To(Equal("https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF/resolve/main/Qwen3VL-2B-Instruct-Q4_K_M.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[0].SHA256).ToNot(BeEmpty(), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[1].Filename).To(Equal("mmproj/mmproj-Qwen3VL-2B-Instruct-Q8_0.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[1].URI).To(Equal("https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF/resolve/main/mmproj-Qwen3VL-2B-Instruct-Q8_0.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[1].SHA256).ToNot(BeEmpty(), fmt.Sprintf("Model config: %+v", modelConfig))
+		})
+
+		It("should discover and import using LlamaCPPImporter", func() {
+			uri := "https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF"
+			preferences := json.RawMessage(`{ "quantizations": "Q8_0", "mmproj_quantizations": "f16" }`)
+
+			modelConfig, err := importers.DiscoverModelConfig(uri, preferences)
+
+			Expect(err).ToNot(HaveOccurred(), fmt.Sprintf("Error: %v", err))
+			Expect(modelConfig.Name).To(Equal("Qwen3-VL-2B-Instruct-GGUF"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Description).To(Equal("Imported from https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.ConfigFile).To(ContainSubstring("backend: llama-cpp"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.ConfigFile).To(ContainSubstring("mmproj: mmproj/mmproj-Qwen3VL-2B-Instruct-F16.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.ConfigFile).To(ContainSubstring("model: Qwen3VL-2B-Instruct-Q8_0.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(len(modelConfig.Files)).To(Equal(2), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[0].Filename).To(Equal("Qwen3VL-2B-Instruct-Q8_0.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[0].URI).To(Equal("https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF/resolve/main/Qwen3VL-2B-Instruct-Q8_0.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[0].SHA256).ToNot(BeEmpty(), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[1].Filename).To(Equal("mmproj/mmproj-Qwen3VL-2B-Instruct-F16.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[1].URI).To(Equal("https://huggingface.co/Qwen/Qwen3-VL-2B-Instruct-GGUF/resolve/main/mmproj-Qwen3VL-2B-Instruct-F16.gguf"), fmt.Sprintf("Model config: %+v", modelConfig))
+			Expect(modelConfig.Files[1].SHA256).ToNot(BeEmpty(), fmt.Sprintf("Model config: %+v", modelConfig))
+		})
 	})
 
 	Context("with .gguf URI", func() {
diff --git a/core/gallery/importers/llama-cpp.go b/core/gallery/importers/llama-cpp.go
index 65bebf7d1966..f0e3b9e596f6 100644
--- a/core/gallery/importers/llama-cpp.go
+++ b/core/gallery/importers/llama-cpp.go
@@ -69,7 +69,7 @@ func (i *LlamaCPPImporter) Import(details Details) (gallery.ModelConfig, error)
 	}
 
 	preferedQuantizations, _ := preferencesMap["quantizations"].(string)
-	quants := []string{"q4_k_m", "q4_0", "q8_0", "f16"}
+	quants := []string{"q4_k_m"}
 	if preferedQuantizations != "" {
 		quants = strings.Split(preferedQuantizations, ",")
 	}
@@ -100,7 +100,7 @@ func (i *LlamaCPPImporter) Import(details Details) (gallery.ModelConfig, error)
 		Description: description,
 	}
 
-	if strings.Contains(details.URI, ".gguf") {
+	if strings.HasSuffix(details.URI, ".gguf") {
 		cfg.Files = append(cfg.Files, gallery.File{
 			URI:      details.URI,
 			Filename: filepath.Base(details.URI),
@@ -111,46 +111,62 @@ func (i *LlamaCPPImporter) Import(details Details) (gallery.ModelConfig, error)
 			},
 		}
 	} else if details.HuggingFace != nil {
-		lastMMProjFile := gallery.File{}
+		// We want to:
+		// Get first the chosen quants that match filenames
+		// OR the first mmproj/gguf file found
+		var lastMMProjFile *gallery.File
+		var lastGGUFFile *gallery.File
 		foundPreferedQuant := false
+		foundPreferedMMprojQuant := false
 
 		for _, file := range details.HuggingFace.Files {
-			// get the files of the prefered quants
-			if slices.ContainsFunc(quants, func(quant string) bool {
-				return strings.Contains(strings.ToLower(file.Path), strings.ToLower(quant))
-			}) {
-				cfg.Files = append(cfg.Files, gallery.File{
-					URI:      file.URL,
-					Filename: filepath.Base(file.Path),
-					SHA256:   file.SHA256,
-				})
-			}
 			// Get the mmproj prefered quants
 			if strings.Contains(strings.ToLower(file.Path), "mmproj") {
-				lastMMProjFile = gallery.File{
+				lastMMProjFile = &gallery.File{
 					URI:      file.URL,
-					Filename: filepath.Base(file.Path),
+					Filename: filepath.Join("mmproj", filepath.Base(file.Path)),
 					SHA256:   file.SHA256,
 				}
 				if slices.ContainsFunc(mmprojQuantsList, func(quant string) bool {
 					return strings.Contains(strings.ToLower(file.Path), strings.ToLower(quant))
+				}) {
+					cfg.Files = append(cfg.Files, *lastMMProjFile)
+					foundPreferedMMprojQuant = true
+				}
+			} else if strings.HasSuffix(strings.ToLower(file.Path), "gguf") {
+				lastGGUFFile = &gallery.File{
+					URI:      file.URL,
+					Filename: filepath.Base(file.Path),
+					SHA256:   file.SHA256,
+				}
+				// get the files of the prefered quants
+				if slices.ContainsFunc(quants, func(quant string) bool {
+					return strings.Contains(strings.ToLower(file.Path), strings.ToLower(quant))
 				}) {
 					foundPreferedQuant = true
-					cfg.Files = append(cfg.Files, lastMMProjFile)
+					cfg.Files = append(cfg.Files, *lastGGUFFile)
 				}
 			}
 		}
 
-		if !foundPreferedQuant && lastMMProjFile.URI != "" {
-			cfg.Files = append(cfg.Files, lastMMProjFile)
-			modelConfig.PredictionOptions = schema.PredictionOptions{
-				BasicModelRequest: schema.BasicModelRequest{
-					Model: lastMMProjFile.Filename,
-				},
+		// Make sure to add at least one file if not already present (which is the latest one)
+		if lastMMProjFile != nil && !foundPreferedMMprojQuant {
+			if !slices.ContainsFunc(cfg.Files, func(f gallery.File) bool {
+				return f.Filename == lastMMProjFile.Filename
+			}) {
+				cfg.Files = append(cfg.Files, *lastMMProjFile)
+			}
+		}
+
+		if lastGGUFFile != nil && !foundPreferedQuant {
+			if !slices.ContainsFunc(cfg.Files, func(f gallery.File) bool {
+				return f.Filename == lastGGUFFile.Filename
+			}) {
+				cfg.Files = append(cfg.Files, *lastGGUFFile)
 			}
 		}
 
-		// Find first mmproj file
+		// Find first mmproj file and configure it in the config file
 		for _, file := range cfg.Files {
 			if !strings.Contains(strings.ToLower(file.Filename), "mmproj") {
 				continue
@@ -159,7 +175,7 @@ func (i *LlamaCPPImporter) Import(details Details) (gallery.ModelConfig, error)
 			break
 		}
 
-		// Find first non-mmproj file
+		// Find first non-mmproj file and configure it in the config file
 		for _, file := range cfg.Files {
 			if strings.Contains(strings.ToLower(file.Filename), "mmproj") {
 				continue