fix(ollama): dedupe latest models during setup

CHANGELOG.md

@@ -21,6 +21,7 @@ Docs: https://docs.openclaw.ai
 ### Fixes
 
 - Media-understanding/audio: migrate deprecated `{input}` placeholders in legacy `audio.transcription.command` configs to `{{MediaPath}}`, so custom audio transcribers no longer receive the literal placeholder after doctor repair. Fixes #72760. Thanks @krisfanue3-hash.
+- Ollama/onboarding: de-dupe suggested bare local models against installed `:latest` tags and skip redundant pulls, so setup shows the installed model once and no longer says it is downloading an already available model. Fixes #68952. Thanks @tleyden.
 - Control UI/Gateway: preserve WebChat client version labels across localhost, 127.0.0.1, and IPv6 loopback aliases on the same port, avoiding misleading `vcontrol-ui` connection logs while investigating duplicate-message reports. Refs #72753 and #72742. Thanks @LumenFromTheFuture and @allesgutefy.
 - Agents/reasoning: treat orphan closing reasoning tags with following answer text as a privacy boundary across delivery, history, streaming, and Control UI sanitizers so malformed local-model output cannot leak chain-of-thought text. Fixes #67092. Thanks @AnildoSilva.
 - Memory-core: run one-shot memory CLI commands through transient builtin and QMD managers so `memory index`, `memory status --index`, and `memory search` no longer start long-lived file watchers that can hit macOS `EMFILE` limits. Fixes #59101; carries forward #49851. Thanks @mbear469210-coder and @maoyuanxue.

docs/providers/ollama.md

@@ -59,7 +59,7 @@ Choose your preferred setup method and mode.
         - **Local only** — local models only
       </Step>
       <Step title="Select a model">
-        `Cloud only` prompts for `OLLAMA_API_KEY` and suggests hosted cloud defaults. `Cloud + Local` and `Local only` ask for an Ollama base URL, discover available models, and auto-pull the selected local model if it is not available yet. `Cloud + Local` also checks whether that Ollama host is signed in for cloud access.
+        `Cloud only` prompts for `OLLAMA_API_KEY` and suggests hosted cloud defaults. `Cloud + Local` and `Local only` ask for an Ollama base URL, discover available models, and auto-pull the selected local model if it is not available yet. When Ollama reports an installed `:latest` tag such as `gemma4:latest`, setup shows that installed model once instead of showing both `gemma4` and `gemma4:latest` or pulling the bare alias again. `Cloud + Local` also checks whether that Ollama host is signed in for cloud access.
       </Step>
       <Step title="Verify the model is available">
         ```bash

extensions/ollama/src/setup.test.ts

@@ -239,6 +239,21 @@ describe("ollama setup", () => {
     expect(modelIds).toContain("llama3:8b");
   });
 
+  it("dedupes the suggested local model against a discovered latest tag", async () => {
+    const prompter = createLocalPrompter();
+
+    const fetchMock = createOllamaFetchMock({ tags: ["gemma4:latest", "llama3:8b"] });
+    vi.stubGlobal("fetch", fetchMock);
+
+    const result = await promptAndConfigureOllama({
+      cfg: {},
+      prompter,
+    });
+
+    const modelIds = result.config.models?.providers?.ollama?.models?.map((m) => m.id);
+    expect(modelIds).toEqual(["gemma4:latest", "llama3:8b"]);
+  });
+
   it("cloud mode does not hit local Ollama endpoints", async () => {
     const prompter = createCloudPrompter();
     const fetchMock = createOllamaFetchMock({ tags: [] });
@@ -536,6 +551,21 @@ describe("ollama setup", () => {
       expect(fetchMock).toHaveBeenCalledTimes(1);
     });
 
+    it("skips pull when an untagged model is available as latest", async () => {
+      const prompter = {} as unknown as WizardPrompter;
+
+      const fetchMock = createOllamaFetchMock({ tags: ["gemma4:latest"] });
+      vi.stubGlobal("fetch", fetchMock);
+
+      await ensureOllamaModelPulled({
+        config: createDefaultOllamaConfig("ollama/gemma4"),
+        model: "ollama/gemma4",
+        prompter,
+      });
+
+      expect(fetchMock).toHaveBeenCalledTimes(1);
+    });
+
     it("uses baseURL alias when checking and pulling models", async () => {
       const progress = { update: vi.fn(), stop: vi.fn() };
       const prompter = {
@@ -656,6 +686,32 @@ describe("ollama setup", () => {
     );
   });
 
+  it("uses the discovered latest tag as the non-interactive default without pulling", async () => {
+    const fetchMock = createOllamaFetchMock({ tags: ["gemma4:latest"] });
+    vi.stubGlobal("fetch", fetchMock);
+    const runtime = createRuntime();
+
+    const result = await configureOllamaNonInteractive({
+      nextConfig: {},
+      opts: {
+        customBaseUrl: "http://127.0.0.1:11434",
+      },
+      runtime,
+    });
+
+    expect(fetchMock).toHaveBeenCalledTimes(2);
+    expect(fetchMock.mock.calls.some((call) => requestUrl(call[0]).endsWith("/api/pull"))).toBe(
+      false,
+    );
+    expect(result.models?.providers?.ollama?.models?.map((model) => model.id)).toEqual([
+      "gemma4:latest",
+    ]);
+    expect(result.agents?.defaults?.model).toEqual(
+      expect.objectContaining({ primary: "ollama/gemma4:latest" }),
+    );
+    expect(runtime.log).toHaveBeenCalledWith("Default Ollama model: gemma4:latest");
+  });
+
   it("accepts cloud models in non-interactive mode without pulling", async () => {
     const fetchMock = createOllamaFetchMock({ tags: [] });
     vi.stubGlobal("fetch", fetchMock);

extensions/ollama/src/setup.ts

@@ -420,21 +420,49 @@ function buildOllamaModelsConfig(
   });
 }
 
+function getOllamaLatestDedupeKey(name: string): string {
+  const normalized = normalizeLowercaseStringOrEmpty(name);
+  return normalized.endsWith(":latest") ? normalized.slice(0, -":latest".length) : normalized;
+}
+
+function isExplicitLatestOllamaModel(name: string): boolean {
+  return normalizeLowercaseStringOrEmpty(name).endsWith(":latest");
+}
+
+function shouldReplaceOllamaModelName(existing: string, candidate: string): boolean {
+  return !isExplicitLatestOllamaModel(existing) && isExplicitLatestOllamaModel(candidate);
+}
+
 function mergeUniqueModelNames(...groups: string[][]): string[] {
-  const seen = new Set<string>();
+  const indexByKey = new Map<string, number>();
   const merged: string[] = [];
   for (const group of groups) {
     for (const name of group) {
-      if (seen.has(name)) {
+      const key = getOllamaLatestDedupeKey(name);
+      const existingIndex = indexByKey.get(key);
+      if (existingIndex !== undefined) {
+        if (shouldReplaceOllamaModelName(merged[existingIndex], name)) {
+          merged[existingIndex] = name;
+        }
         continue;
       }
-      seen.add(name);
+      indexByKey.set(key, merged.length);
       merged.push(name);
     }
   }
   return merged;
 }
 
+function findAvailableOllamaModelName(modelName: string, availableModelNames: Iterable<string>) {
+  const wantedKey = getOllamaLatestDedupeKey(modelName);
+  for (const available of availableModelNames) {
+    if (getOllamaLatestDedupeKey(available) === wantedKey) {
+      return available;
+    }
+  }
+  return undefined;
+}
+
 function applyOllamaProviderConfig(
   cfg: OpenClawConfig,
   baseUrl: string,
@@ -632,19 +660,20 @@ export async function configureOllamaNonInteractive(params: {
   );
   const discoveredModelsByName = new Map(enrichedModels.map((model) => [model.name, model]));
   const modelNames = models.map((model) => model.name);
-  const orderedModelNames = [
-    ...OLLAMA_SUGGESTED_MODELS_LOCAL,
-    ...modelNames.filter((name) => !OLLAMA_SUGGESTED_MODELS_LOCAL.includes(name)),
-  ];
+  const orderedModelNames = mergeUniqueModelNames(OLLAMA_SUGGESTED_MODELS_LOCAL, modelNames);
 
   const requestedDefaultModelId = explicitModel ?? OLLAMA_SUGGESTED_MODELS_LOCAL[0];
   const availableModelNames = new Set(modelNames);
+  const availableDefaultModelId = findAvailableOllamaModelName(
+    requestedDefaultModelId,
+    availableModelNames,
+  );
   const requestedCloudModel = isOllamaCloudModel(requestedDefaultModelId);
   let pulledRequestedModel = false;
 
   if (requestedCloudModel) {
     availableModelNames.add(requestedDefaultModelId);
-  } else if (!modelNames.includes(requestedDefaultModelId)) {
+  } else if (!availableDefaultModelId) {
     pulledRequestedModel = await pullOllamaModelNonInteractive(
       baseUrl,
       requestedDefaultModelId,
@@ -656,7 +685,7 @@ export async function configureOllamaNonInteractive(params: {
   }
 
   let allModelNames = orderedModelNames;
-  let defaultModelId = requestedDefaultModelId;
+  let defaultModelId = availableDefaultModelId ?? requestedDefaultModelId;
   if (
     (pulledRequestedModel || requestedCloudModel) &&
     !allModelNames.includes(requestedDefaultModelId)
@@ -664,7 +693,7 @@ export async function configureOllamaNonInteractive(params: {
     allModelNames = [...allModelNames, requestedDefaultModelId];
   }
 
-  if (!availableModelNames.has(requestedDefaultModelId)) {
+  if (!findAvailableOllamaModelName(defaultModelId, availableModelNames)) {
     if (availableModelNames.size === 0) {
       params.runtime.error(
         [
@@ -677,7 +706,7 @@ export async function configureOllamaNonInteractive(params: {
     }
 
     defaultModelId =
-      allModelNames.find((name) => availableModelNames.has(name)) ??
+      allModelNames.find((name) => findAvailableOllamaModelName(name, availableModelNames)) ??
       Array.from(availableModelNames)[0];
     params.runtime.log(
       `Ollama model ${requestedDefaultModelId} was not available; using ${defaultModelId} instead.`,
@@ -709,7 +738,12 @@ export async function ensureOllamaModelPulled(params: {
     return;
   }
   const { models } = await fetchOllamaModels(baseUrl);
-  if (models.some((model) => model.name === modelName)) {
+  if (
+    findAvailableOllamaModelName(
+      modelName,
+      models.map((model) => model.name),
+    )
+  ) {
     return;
   }
   if (!(await pullOllamaModel(baseUrl, modelName, params.prompter))) {