fix(ollama): add Gemma to tool support heuristic for Ollama and LM Studio

core/llm/toolSupport.test.ts

@@ -260,6 +260,13 @@ describe("PROVIDER_TOOL_SUPPORT", () => {
       expect(supportsFn("devstral-24b")).toBe(true);
     });
 
+    it("should return true for Gemma models", () => {
+      expect(supportsFn("gemma3")).toBe(true);
+      expect(supportsFn("gemma4")).toBe(true);
+      expect(supportsFn("gemma3:27b")).toBe(true);
+      expect(supportsFn("gemma4:27b")).toBe(true);
+    });
+
     it("should return false for explicitly unsupported models", () => {
       expect(supportsFn("vision")).toBe(false);
       expect(supportsFn("math")).toBe(false);
@@ -292,6 +299,14 @@ describe("PROVIDER_TOOL_SUPPORT", () => {
       expect(supportsFn("mistral-7b")).toBe(true);
     });
 
+    it("should return true for Gemma models via LM Studio", () => {
+      // Gemma 3/4 support function calling per Google's documentation
+      expect(supportsFn("gemma3")).toBe(true);
+      expect(supportsFn("gemma4")).toBe(true);
+      expect(supportsFn("Gemma-3-27B-Instruct-GGUF")).toBe(true);
+      expect(supportsFn("Gemma-4-27B-Instruct-GGUF")).toBe(true);
+    });
+
     it("should return true for LM Studio hyphenated model IDs", () => {
       // LM Studio uses hyphenated model identifiers like "Meta-Llama-3.1-8B-Instruct-GGUF"
       expect(supportsFn("Meta-Llama-3.1-8B-Instruct-GGUF")).toBe(true);

core/llm/toolSupport.ts

@@ -213,6 +213,8 @@ export const PROVIDER_TOOL_SUPPORT: Record<string, (model: string) => boolean> =
           "glm-5",
           "deepseek",
           "dolphin",
+          // https://ai.google.dev/gemma/docs/capabilities/function-calling
+          "gemma",
         ].some((part) => modelName.toLowerCase().includes(part))
       ) {
         return true;