diff --git a/src/commands/providers.ts b/src/commands/providers.ts
index 68c22c7a2..5d31d4c68 100644
--- a/src/commands/providers.ts
+++ b/src/commands/providers.ts
@@ -7,9 +7,9 @@
 
 import { listRecipes, getRecipe } from '../core/ai/recipes/index.ts';
 import { configureGateway, embedOne, isAvailable as gwIsAvailable, chat as gwChat } from '../core/ai/gateway.ts';
+import { buildGatewayConfig } from '../core/ai/build-gateway-config.ts';
 import { probeOllama, probeLMStudio } from '../core/ai/probes.ts';
 import { loadConfig } from '../core/config.ts';
-import { buildGatewayConfig } from '../core/ai/build-gateway-config.ts';
 import { AIConfigError, AITransientError } from '../core/ai/errors.ts';
 import type { Recipe } from '../core/ai/types.ts';
 
@@ -173,8 +173,18 @@ async function runTest(args: string[]): Promise<void> {
     // the divergence at the top of the test so the recovery experience
     // doesn't repeat the bug-reporter's "providers test ✓ but import still
     // broken" trap.
+    //
+    // #2863: `cfg` is lifted out of the try block (not just used for the
+    // warning) so the configureGateway calls below can reuse it. Before this
+    // fix, the --model override only forwarded embedding_model/chat_model +
+    // env, dropping config.provider_base_urls entirely — a probe against a
+    // custom endpoint (e.g. a regional DashScope base URL) would silently
+    // fall back to the recipe's hardcoded default endpoint and fail with a
+    // misleading "Incorrect API key" error even though the key was valid for
+    // the configured endpoint.
+    let cfg: ReturnType<typeof loadConfig> | null = null;
     try {
-      const cfg = loadConfig();
+      cfg = loadConfig();
       const configuredModel = tpArg === 'embedding' ? cfg?.embedding_model : cfg?.chat_model;
       if (!configuredModel) {
         console.error(
@@ -190,17 +200,27 @@ async function runTest(args: string[]): Promise<void> {
       }
     } catch { /* loadConfig throws when no brain configured — first-time install path; the no-config branch above handles it. */ }
 
+    // Reuse the SAME resolver the production path uses (buildGatewayConfig —
+    // also used by cli.ts#connectEngine and init-embed-check.ts) so the probe
+    // sees the identical base_urls / provider_chat_options / folded API keys
+    // that a real `gbrain import`/`gbrain query` call would. Only the
+    // touchpoint's model (+ embedding dims) is overridden on top, so an
+    // isolated `--model` probe still targets exactly the requested model —
+    // it just resolves that model's endpoint the way the brain actually
+    // would. Falls back to bare env when no brain is configured yet (cfg is
+    // null on first-time install, matching the old behavior for that case).
+    const baseGatewayConfig = cfg ? buildGatewayConfig(cfg) : { env: { ...process.env } };
     if (tpArg === 'embedding') {
       const dims = recipe?.touchpoints.embedding?.default_dims ?? 1536;
       configureGateway({
+        ...baseGatewayConfig,
         embedding_model: modelArg,
         embedding_dimensions: dims,
-        env: { ...process.env },
       });
     } else {
       configureGateway({
+        ...baseGatewayConfig,
         chat_model: modelArg,
-        env: { ...process.env },
       });
     }
     void modelId; // intentionally unused but preserved for readability
diff --git a/test/providers-test-model-base-url.test.ts b/test/providers-test-model-base-url.test.ts
new file mode 100644
index 000000000..9442d3ad6
--- /dev/null
+++ b/test/providers-test-model-base-url.test.ts
@@ -0,0 +1,107 @@
+/**
+ * #2863 regression — `gbrain providers test --model` must resolve
+ * `provider_base_urls` the same way the production embed/chat path does.
+ *
+ * Before the fix, the `--model` override branch in `runTest`
+ * (src/commands/providers.ts) forwarded only `embedding_model`/`chat_model`
+ * + `env` into `configureGateway`, dropping `config.provider_base_urls`
+ * entirely. A brain configured with a custom (e.g. China-region DashScope)
+ * endpoint would pass `gbrain providers test --touchpoint embedding` (no
+ * `--model`, uses configureFromEnv() which DOES forward base_urls) but fail
+ * `gbrain providers test --touchpoint embedding --model
+ * dashscope:text-embedding-v3` with a misleading "Incorrect API key" error
+ * — the probe silently fell back to the recipe's hardcoded default endpoint
+ * (dashscope-intl.aliyuncs.com) instead of the configured one.
+ *
+ * This test drives the real `runProviders('test', ...)` CLI path end to end
+ * (loadConfig -> configureGateway -> gateway -> AI SDK -> fetch) and asserts
+ * on the actual HTTP request URL, so it fails on the pre-fix code and only
+ * passes once the --model override reuses buildGatewayConfig (the same
+ * resolver src/cli.ts#connectEngine and init-embed-check.ts use for the
+ * production path).
+ */
+
+import { afterEach, beforeEach, describe, expect, test } from 'bun:test';
+import { mkdtempSync, mkdirSync, writeFileSync, rmSync } from 'node:fs';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { runProviders } from '../src/commands/providers.ts';
+import { resetGateway } from '../src/core/ai/gateway.ts';
+import { withEnv } from './helpers/with-env.ts';
+
+const CUSTOM_BASE_URL = 'https://llm-custom.cn-beijing.maas.example.test/compatible-mode/v1';
+
+type FetchHandler = (url: string, init: RequestInit) => Promise<Response>;
+let fetchHandler: FetchHandler | null = null;
+const origFetch = globalThis.fetch;
+let tmpHome: string;
+
+function okEmbeddingResponse(dims: number): Response {
+  const vec = Array(dims).fill(0).map((_, i) => 0.001 * i);
+  return new Response(
+    JSON.stringify({ data: [{ embedding: vec }] }),
+    { status: 200, headers: { 'Content-Type': 'application/json' } },
+  );
+}
+
+beforeEach(() => {
+  fetchHandler = null;
+  globalThis.fetch = (async (url: string | URL | Request, init?: RequestInit) => {
+    if (!fetchHandler) throw new Error('fetch called but no handler installed');
+    return fetchHandler(typeof url === 'string' ? url : url.toString(), init ?? {});
+  }) as typeof fetch;
+
+  tmpHome = mkdtempSync(join(tmpdir(), 'gbrain-providers-test-base-url-'));
+  mkdirSync(join(tmpHome, '.gbrain'), { recursive: true });
+  writeFileSync(
+    join(tmpHome, '.gbrain', 'config.json'),
+    JSON.stringify({
+      embedding_model: 'dashscope:text-embedding-v3',
+      embedding_dimensions: 1024,
+      provider_base_urls: { dashscope: CUSTOM_BASE_URL },
+    }),
+  );
+});
+
+afterEach(() => {
+  globalThis.fetch = origFetch;
+  resetGateway();
+  rmSync(tmpHome, { recursive: true, force: true });
+});
+
+describe('providers test --model — provider_base_urls (#2863)', () => {
+  test('embedding touchpoint probe hits the configured custom base URL, not the recipe default', async () => {
+    let capturedUrl = '';
+    fetchHandler = async (url) => {
+      capturedUrl = url;
+      return okEmbeddingResponse(1024);
+    };
+
+    await withEnv(
+      { GBRAIN_HOME: tmpHome, DASHSCOPE_API_KEY: 'test-dashscope-key' },
+      async () => {
+        await runProviders('test', ['--touchpoint', 'embedding', '--model', 'dashscope:text-embedding-v3']);
+      },
+    );
+
+    expect(capturedUrl.startsWith(CUSTOM_BASE_URL)).toBe(true);
+    expect(capturedUrl).not.toContain('dashscope-intl.aliyuncs.com');
+  });
+
+  test('bare `providers test` (no --model) already used the custom base URL (control)', async () => {
+    let capturedUrl = '';
+    fetchHandler = async (url) => {
+      capturedUrl = url;
+      return okEmbeddingResponse(1024);
+    };
+
+    await withEnv(
+      { GBRAIN_HOME: tmpHome, DASHSCOPE_API_KEY: 'test-dashscope-key' },
+      async () => {
+        await runProviders('test', ['--touchpoint', 'embedding']);
+      },
+    );
+
+    expect(capturedUrl.startsWith(CUSTOM_BASE_URL)).toBe(true);
+  });
+});