diff --git a/open-sse/providers/registry/openrouter.js b/open-sse/providers/registry/openrouter.js
index 4ac036411..c454d2ed9 100644
--- a/open-sse/providers/registry/openrouter.js
+++ b/open-sse/providers/registry/openrouter.js
@@ -38,8 +38,16 @@ export default {
     { id: "openai/gpt-image-1", name: "GPT Image 1 (via OpenRouter)", params: ["n","size","quality","response_format"], kind: "image" },
     { id: "google/imagen-3.0-generate-002", name: "Imagen 3 (via OpenRouter)", params: ["n","size"], kind: "image" },
     { id: "black-forest-labs/FLUX.1-schnell", name: "FLUX.1 Schnell (via OpenRouter)", params: ["n","size"], kind: "image" },
+    // OpenRouter exposes a separate, Cohere-compatible POST /api/v1/rerank endpoint
+    // (not surfaced by its live /v1/models feed). Model IDs keep their vendor slash
+    // (e.g. "cohere/rerank-4-pro"); the model parser splits on the first slash, so
+    // 3-segment ids resolve safely. Seeded by hand as OpenRouter adds rerank models.
+    { id: "cohere/rerank-4-pro", name: "Cohere Rerank 4 Pro (via OpenRouter)", kind: "rerank" },
+    { id: "cohere/rerank-4-fast", name: "Cohere Rerank 4 Fast (via OpenRouter)", kind: "rerank" },
+    { id: "cohere/rerank-v3.5", name: "Cohere Rerank v3.5 (via OpenRouter)", kind: "rerank" },
+    { id: "nvidia/llama-nemotron-rerank-vl-1b-v2:free", name: "Llama Nemotron Rerank VL 1B v2 (free, via OpenRouter)", kind: "rerank" },
   ],
-  serviceKinds: ["llm","embedding","tts","imageToText"],
+  serviceKinds: ["llm","embedding","tts","imageToText","rerank"],
   ttsConfig: {
     baseUrl: "https://openrouter.ai/api/v1/chat/completions",
     defaultModel: "openai/gpt-4o-mini-tts",
diff --git a/src/app/api/models/test/ping.js b/src/app/api/models/test/ping.js
index 0a2cc9830..80ae927d6 100644
--- a/src/app/api/models/test/ping.js
+++ b/src/app/api/models/test/ping.js
@@ -122,6 +122,37 @@ export async function pingModelByKind(model, kind, baseUrl = `http://127.0.0.1:$
     return { ok: true, latencyMs, error: null, status: res.status };
   }
 
+  if (kind === "rerank") {
+    const res = await fetch(`${baseUrl}/api/v1/rerank`, {
+      method: "POST",
+      headers,
+      body: JSON.stringify({
+        model,
+        query: "ping",
+        documents: ["hello world"],
+        top_n: 1,
+      }),
+      signal: AbortSignal.timeout(15000),
+    });
+    const latencyMs = Date.now() - start;
+    const rawText = await res.text().catch(() => "");
+    let parsed = null;
+    try { parsed = rawText ? JSON.parse(rawText) : null; } catch {}
+
+    if (!res.ok) {
+      const detail = parsed?.error?.message || parsed?.msg || parsed?.message || parsed?.error || rawText;
+      return { ok: false, latencyMs, error: `HTTP ${res.status}${detail ? `: ${String(detail).slice(0, 240)}` : ""}`, status: res.status };
+    }
+
+    const results = Array.isArray(parsed?.results) ? parsed.results
+      : Array.isArray(parsed?.data) ? parsed.data
+        : null;
+    if (!results) {
+      return { ok: false, latencyMs, status: res.status, error: "Provider returned no rerank results for this model" };
+    }
+    return { ok: true, latencyMs, error: null, status: res.status };
+  }
+
   const res = await fetch(`${baseUrl}/api/v1/chat/completions`, {
     method: "POST",
     headers,
diff --git a/src/app/api/v1/models/[kind]/route.js b/src/app/api/v1/models/[kind]/route.js
index d6a649a5c..c015278ca 100644
--- a/src/app/api/v1/models/[kind]/route.js
+++ b/src/app/api/v1/models/[kind]/route.js
@@ -10,6 +10,7 @@ const KIND_SLUG_MAP = {
   "embedding": ["embedding"],
   "image-to-text": ["imageToText"],
   "web": ["webSearch", "webFetch"],
+  "rerank": ["rerank"],
 };
 
 export async function OPTIONS() {
@@ -24,7 +25,7 @@ export async function OPTIONS() {
 
 /**
  * GET /v1/models/{kind} - OpenAI-compatible models list filtered by capability.
- * Supported kinds: image, tts, stt, embedding, image-to-text, web.
+ * Supported kinds: image, tts, stt, embedding, image-to-text, web, rerank.
  */
 export async function GET(request, { params }) {
   try {
diff --git a/src/app/api/v1/models/buildModelsList.js b/src/app/api/v1/models/buildModelsList.js
index 6e4c773eb..fb031bd51 100644
--- a/src/app/api/v1/models/buildModelsList.js
+++ b/src/app/api/v1/models/buildModelsList.js
@@ -187,6 +187,7 @@ const MODEL_TYPE_TO_KIND = {
   embedding: "embedding",
   stt: "stt",
   imageToText: "imageToText",
+  rerank: "rerank",
 };
 
 function modelKind(model) {
diff --git a/src/app/api/v1/models/info/route.js b/src/app/api/v1/models/info/route.js
index f659fa357..9c4971894 100644
--- a/src/app/api/v1/models/info/route.js
+++ b/src/app/api/v1/models/info/route.js
@@ -12,6 +12,7 @@ const KIND_ENDPOINT = {
   imageToText: "/v1/chat/completions",
   webSearch: "/v1/search",
   webFetch: "/v1/fetch",
+  rerank: "/v1/rerank",
 };
 
 const TTS_VOICES_API = new Set(["elevenlabs", "edge-tts", "deepgram", "inworld", "local-device", "minimax", "minimax-cn"]);
diff --git a/src/shared/constants/providers.js b/src/shared/constants/providers.js
index 8678e5bf2..d42a1077d 100644
--- a/src/shared/constants/providers.js
+++ b/src/shared/constants/providers.js
@@ -75,6 +75,7 @@ export const WEB_COOKIE_PROVIDERS = byCategory("webCookie");
 // Media provider kinds — each kind maps to a route and endpoint config
 export const MEDIA_PROVIDER_KINDS = [
   { id: "embedding",   label: "Embedding",      icon: "data_array",        endpoint: { method: "POST", path: "/v1/embeddings" } },
+  { id: "rerank",      label: "Rerank",         icon: "sort",              endpoint: { method: "POST", path: "/v1/rerank" } },
   { id: "image",       label: "Text to Image",  icon: "brush",             endpoint: { method: "POST", path: "/v1/images/generations" } },
   { id: "imageToText", label: "Image to Text",  icon: "image_search",      endpoint: { method: "POST", path: "/v1/images/understanding" } },
   { id: "tts",         label: "Text To Speech", icon: "record_voice_over", endpoint: { method: "POST", path: "/v1/audio/speech" } },
diff --git a/tests/unit/models-info-rerank-endpoint.test.js b/tests/unit/models-info-rerank-endpoint.test.js
new file mode 100644
index 000000000..efb3e8638
--- /dev/null
+++ b/tests/unit/models-info-rerank-endpoint.test.js
@@ -0,0 +1,18 @@
+import { describe, expect, it } from "vitest";
+
+// PR #139 review (Codex): /v1/models/rerank advertises openrouter/cohere/rerank-*
+// but GET /v1/models/info built endpoint metadata from KIND_ENDPOINT which had
+// no `rerank` entry, so the info response returned endpoint: null and clients
+// could not route the newly-discoverable rerank models. Pin the endpoint.
+describe("GET /v1/models/info rerank endpoint metadata", () => {
+  it("returns endpoint /v1/rerank for a rerank model", async () => {
+    const { GET } = await import("../../src/app/api/v1/models/info/route.js");
+    const res = await GET(
+      new Request("http://local.test/v1/models/info?id=openrouter/cohere/rerank-4-pro"),
+    );
+    expect(res.status).toBe(200);
+    const body = await res.json();
+    expect(body.kind).toBe("rerank");
+    expect(body.endpoint).toBe("/v1/rerank");
+  });
+});
diff --git a/tests/unit/openrouter-rerank-buildModelsList.test.js b/tests/unit/openrouter-rerank-buildModelsList.test.js
new file mode 100644
index 000000000..8724cc128
--- /dev/null
+++ b/tests/unit/openrouter-rerank-buildModelsList.test.js
@@ -0,0 +1,62 @@
+import { beforeEach, describe, expect, it, vi } from "vitest";
+
+// PR #139 review (Codex): static OpenRouter rerank rows carry `kind: "rerank"`.
+// Without `buildModelsList` mapping that kind, the LLM filter would advertise
+// rerank-only models as chat models. This test pins both halves of the fix:
+//   - /v1/models (LLM filter) MUST NOT surface openrouter rerank models
+//   - /v1/models/rerank (rerank filter) MUST surface them
+const mocks = vi.hoisted(() => ({
+  getProviderConnections: vi.fn(),
+  getCombos: vi.fn(),
+  getCustomModels: vi.fn(),
+  getModelAliases: vi.fn(),
+  getDisabledModels: vi.fn(),
+}));
+
+vi.mock("@/lib/localDb", () => ({
+  getProviderConnections: mocks.getProviderConnections,
+  getCombos: mocks.getCombos,
+  getCustomModels: mocks.getCustomModels,
+  getModelAliases: mocks.getModelAliases,
+}));
+
+vi.mock("@/lib/disabledModelsDb", () => ({
+  getDisabledModels: mocks.getDisabledModels,
+}));
+
+describe("buildModelsList rerank kind handling", () => {
+  beforeEach(() => {
+    vi.clearAllMocks();
+    mocks.getProviderConnections.mockResolvedValue([
+      { provider: "openrouter", isActive: true, apiKey: "sk-test" },
+    ]);
+    mocks.getCombos.mockResolvedValue([]);
+    mocks.getCustomModels.mockResolvedValue([]);
+    mocks.getModelAliases.mockResolvedValue({});
+    mocks.getDisabledModels.mockResolvedValue({});
+  });
+
+  it("excludes rerank-only models from the chat-completions (llm) list", async () => {
+    const { buildModelsList, LLM_KIND } = await import(
+      "../../src/app/api/v1/models/buildModelsList.js"
+    );
+    const models = await buildModelsList([LLM_KIND]);
+    const ids = new Set(models.map((m) => m.id));
+
+    expect(ids.has("openrouter/cohere/rerank-4-pro")).toBe(false);
+    expect(ids.has("openrouter/cohere/rerank-4-fast")).toBe(false);
+    expect(ids.has("openrouter/cohere/rerank-v3.5")).toBe(false);
+    expect(ids.has("openrouter/nvidia/llama-nemotron-rerank-vl-1b-v2:free")).toBe(false);
+  });
+
+  it("surfaces rerank-only models under the rerank kind filter", async () => {
+    const { buildModelsList } = await import(
+      "../../src/app/api/v1/models/buildModelsList.js"
+    );
+    const models = await buildModelsList(["rerank"]);
+    const ids = new Set(models.map((m) => m.id));
+
+    expect(ids.has("openrouter/cohere/rerank-4-pro")).toBe(true);
+    expect(ids.has("openrouter/nvidia/llama-nemotron-rerank-vl-1b-v2:free")).toBe(true);
+  });
+});
diff --git a/tests/unit/ping-rerank-endpoint.test.js b/tests/unit/ping-rerank-endpoint.test.js
new file mode 100644
index 000000000..917667ce3
--- /dev/null
+++ b/tests/unit/ping-rerank-endpoint.test.js
@@ -0,0 +1,69 @@
+import { afterEach, beforeEach, describe, expect, it, vi } from "vitest";
+
+// PR #139 review (Codex): pingModelByKind used to fall every non-embedding/
+// image/stt kind through to POST /api/v1/chat/completions, so testing a
+// rerank-only model reported a chat-completion failure. Pin the new rerank
+// branch: it MUST hit /api/v1/rerank and MUST NOT hit chat completions.
+const mocks = vi.hoisted(() => ({
+  getApiKeys: vi.fn(),
+  getConsistentMachineId: vi.fn(),
+}));
+
+vi.mock("@/lib/localDb", () => ({
+  getApiKeys: mocks.getApiKeys,
+}));
+
+vi.mock("@/shared/utils/machineId", () => ({
+  getConsistentMachineId: mocks.getConsistentMachineId,
+}));
+
+const originalFetch = global.fetch;
+
+describe("pingModelByKind rerank endpoint", () => {
+  let calls;
+
+  beforeEach(() => {
+    vi.resetModules();
+    mocks.getApiKeys.mockResolvedValue([]);
+    mocks.getConsistentMachineId.mockResolvedValue("machine-id-test");
+    calls = [];
+    global.fetch = vi.fn(async (url) => {
+      calls.push(String(url));
+      return {
+        ok: true,
+        status: 200,
+        text: async () => JSON.stringify({ results: [{ index: 0, relevance_score: 0.9 }] }),
+      };
+    });
+  });
+
+  afterEach(() => {
+    global.fetch = originalFetch;
+  });
+
+  it("calls POST /api/v1/rerank, not /api/v1/chat/completions", async () => {
+    const { pingModelByKind } = await import(
+      "../../src/app/api/models/test/ping.js"
+    );
+    const result = await pingModelByKind(
+      "openrouter/cohere/rerank-4-pro",
+      "rerank",
+      "http://local.test",
+    );
+
+    expect(result.ok).toBe(true);
+    expect(calls.some((u) => u === "http://local.test/api/v1/rerank")).toBe(true);
+    expect(calls.some((u) => u.endsWith("/api/v1/chat/completions"))).toBe(false);
+
+    const [url, init] = global.fetch.mock.calls[0];
+    expect(url).toBe("http://local.test/api/v1/rerank");
+    expect(init.method).toBe("POST");
+    const body = JSON.parse(init.body);
+    expect(body).toMatchObject({
+      model: "openrouter/cohere/rerank-4-pro",
+      query: "ping",
+      documents: ["hello world"],
+      top_n: 1,
+    });
+  });
+});
diff --git a/tests/unit/rerank-openrouter-6574.test.js b/tests/unit/rerank-openrouter-6574.test.js
new file mode 100644
index 000000000..4c7bd0820
--- /dev/null
+++ b/tests/unit/rerank-openrouter-6574.test.js
@@ -0,0 +1,42 @@
+import { describe, it, expect } from "vitest";
+import {
+  PROVIDERS,
+  PROVIDER_MEDIA,
+  PROVIDER_MODELS,
+} from "open-sse/providers/index.js";
+import { deriveRerankUrl } from "open-sse/handlers/rerankCore.js";
+import { parseModel } from "open-sse/services/model.js";
+import openrouter from "open-sse/providers/registry/openrouter.js";
+
+// #6574 — OpenRouter exposes a Cohere-compatible POST /api/v1/rerank endpoint
+// (confirmed live: openrouter.ai/cohere/rerank-4-pro). The capability/model catalog
+// missed rerank registration, so clients could not discover or route OpenRouter rerank
+// models reliably.
+describe("openrouter rerank provider registration", () => {
+  it("parseModel resolves openrouter multi-slash rerank model id", () => {
+    expect(parseModel("openrouter/cohere/rerank-4-pro")).toEqual({
+      provider: "openrouter",
+      model: "cohere/rerank-4-pro",
+      isAlias: false,
+      providerAlias: "openrouter",
+    });
+  });
+  it("openrouter registry includes rerank service kind", () => {
+    expect(openrouter.serviceKinds).toContain("rerank");
+  });
+
+  it("PROVIDER_MODELS.openrouter contains the rerank model ids", () => {
+    const ids = new Set(PROVIDER_MODELS.openrouter.map((m) => m.id));
+    expect(ids.has("cohere/rerank-4-pro")).toBe(true);
+    expect(ids.has("cohere/rerank-4-fast")).toBe(true);
+    expect(ids.has("cohere/rerank-v3.5")).toBe(true);
+
+    const rerank = PROVIDER_MODELS.openrouter.find((m) => m.id === "cohere/rerank-4-pro");
+    expect(rerank?.kind).toBe("rerank");
+  });
+
+  it("deriveRerankUrl resolves openrouter to its Cohere-compatible /rerank endpoint", () => {
+    const url = deriveRerankUrl(PROVIDERS.openrouter, PROVIDER_MEDIA.openrouter);
+    expect(url).toBe("https://openrouter.ai/api/v1/rerank");
+  });
+});