apps/desktop/e2e/live-reginald.spec.ts

@@ -11,13 +11,13 @@ const MAIN = join(here, '..', 'out', 'main', 'index.js')
 test.describe('live Reginald', () => {
   test('answers a question by consulting the real project', async () => {
     test.skip(!process.env.GITEA_LIVE || !process.env.COMMITEA_MODEL_LIVE, 'live model test — opt-in')
-    test.setTimeout(120_000)
+    test.setTimeout(300_000) // a big local model is slow: ~2 calls/turn + a reconcile
     const app = await electron.launch({ args: [MAIN], env: { ...process.env } })
     const win = await app.firstWindow()
     await win.waitForLoadState('domcontentloaded')
 
-    // model configured → the live greeting + header, not the scripted demo
+    // model configured → the live greeting + header (the loaded model, not the scripted demo)
-    await expect(win.getByText('gemma-4 · local')).toBeVisible({ timeout: 20000 })
+    await expect(win.getByText(/· local$/)).toBeVisible({ timeout: 20000 })
     await expect(win.getByText(/I check the real board before I answer/)).toBeVisible()
 
     const composer = win.getByPlaceholder(/Tell me what to do/)
@@ -25,7 +25,7 @@ test.describe('live Reginald', () => {
     await composer.press('Enter')
 
     // the agent loop ran end-to-end: it consulted the project, then answered
-    await expect(win.getByText(/consulted the project/)).toBeVisible({ timeout: 90_000 })
+    await expect(win.getByText(/consulted the project/)).toBeVisible({ timeout: 240_000 })
     await win.screenshot({ path: join(here, '.artifacts', 'screens', 'live-reginald.png'), fullPage: true, animations: 'disabled' })
 
     await app.close()

apps/desktop/src/main/model.ts

@@ -28,23 +28,57 @@ function resolveModelRouter(): ModelRouter | null {
   const baseUrl = process.env.COMMITEA_MODEL_URL ?? 'http://localhost:1234/v1'
   if (!baseUrl) return null
   return {
-    small: { baseUrl, model: process.env.COMMITEA_MODEL_SMALL ?? 'google/gemma-4-e4b' },
+    small: { baseUrl, model: process.env.COMMITEA_MODEL_SMALL ?? '' },
-    big: { baseUrl, model: process.env.COMMITEA_MODEL_BIG ?? 'qwen/qwen3.6-35b-a3b' },
+    big: { baseUrl, model: process.env.COMMITEA_MODEL_BIG ?? '' },
   }
 }
 
+/**
+ * Resolve which model to actually ask for. An explicit env override wins;
+ * otherwise ask the server which model is *loaded* (LM Studio's native
+ * `/api/v0/models`) so Reginald follows whatever you load — no config churn on a
+ * model switch. Falls back to the first non-embedding model, then a sane default.
+ */
+async function resolveLoadedModel(baseUrl: string, override: string): Promise<string> {
+  if (override) return override
+  const root = baseUrl.replace(/\/v1\/?$/, '')
+  try {
+    const res = await fetch(`${root}/api/v0/models`)
+    if (res.ok) {
+      const data = (await res.json()) as { data?: { id: string; state?: string; type?: string }[] }
+      const loaded = (data.data ?? []).find((m) => m.state === 'loaded' && m.type !== 'embeddings')
+      if (loaded) return loaded.id
+    }
+  } catch {
+    // native API unavailable — fall through to the OpenAI listing
+  }
+  try {
+    const res = await fetch(`${baseUrl.replace(/\/+$/, '')}/models`)
+    if (res.ok) {
+      const data = (await res.json()) as { data?: { id: string }[] }
+      const first = (data.data ?? []).find((m) => !/embed/i.test(m.id))
+      if (first) return first.id
+    }
+  } catch {
+    // ignore — use the default
+  }
+  return 'google/gemma-4-e4b'
+}
+
 export function registerModelIpc(): void {
   const router = resolveModelRouter()
 
-  ipcMain.handle('model:status', () => ({
+  ipcMain.handle('model:status', async () => {
-    configured: !!router,
+    if (!router) return { configured: false, model: null }
-    model: router?.small.model ?? null,
+    const model = await resolveLoadedModel(router.small.baseUrl, router.small.model)
-  }))
+    return { configured: true, model }
+  })
 
   ipcMain.handle('model:chat', async (_event, messages: ChatMessage[]) => {
     if (!router) return { ok: false as const, reason: 'unconfigured' as const }
     const client = getGiteaClient()
-    const chat = createChatClient(router.small, fetch)
+    const model = await resolveLoadedModel(router.small.baseUrl, router.small.model)
+    const chat = createChatClient({ ...router.small, model }, fetch)
 
     const execute = async (name: string, args: unknown) => {
       if (name !== 'query_project') return { error: `unknown tool: ${name}` }

apps/desktop/src/renderer/src/components/shell/chat-panel.tsx

@@ -15,7 +15,9 @@ export interface ChatPanelProps {
 }
 
 export function ChatPanel({ onOpenDirectives, offline }: ChatPanelProps) {
-  const { msgs, thinking, live, steps, send: sendChat } = useChat()
+  const { msgs, thinking, live, model, steps, send: sendChat } = useChat()
+  // shorten "google/gemma-4-26b-a4b-qat" → "gemma-4-26b" for the header chip
+  const modelLabel = model ? (model.split('/').pop() ?? model).replace(/-(qat|instruct|it|gguf)$/i, '') : 'gemma-4'
   const [text, setText] = useState('')
   const scrollRef = useRef<HTMLDivElement>(null)
 
@@ -56,7 +58,7 @@ export function ChatPanel({ onOpenDirectives, offline }: ChatPanelProps) {
         <Icon name="sparkles" size={16} style={{ color: offline ? 'var(--ink-3)' : 'var(--jade)' }} />
         <span style={{ font: 'var(--text-body-strong)', color: 'var(--ink-1)' }}>Reginald</span>
         <span style={{ font: '400 11px var(--font-mono)', color: 'var(--ink-3)', marginLeft: 'auto' }}>
-          {offline ? 'offline · queueing' : live ? 'gemma-4 · local' : 'demo · scripted'}
+          {offline ? 'offline · queueing' : live ? `${modelLabel} · local` : 'demo · scripted'}
         </span>
         <IconButton icon="history" label="Directive log" size="sm" onClick={onOpenDirectives} />
       </header>

apps/desktop/src/renderer/src/lib/use-chat.ts

@@ -14,6 +14,8 @@ export interface ChatState {
   thinking: boolean
   /** true once a model endpoint is confirmed; otherwise the panel echoes the demo reply. */
   live: boolean
+  /** The loaded model's id when live (for the header). */
+  model: string | null
   /** Tools Reginald consulted on the last turn (for a subtle activity line). */
   steps: string[]
   send: (text: string) => void
@@ -31,6 +33,7 @@ export function useChat(): ChatState {
   const [convo, setConvo] = useState<ChatMessage[]>([])
   const [thinking, setThinking] = useState(false)
   const [live, setLive] = useState(false)
+  const [model, setModel] = useState<string | null>(null)
   const [steps, setSteps] = useState<string[]>([])
   const convoRef = useRef(convo)
   convoRef.current = convo
@@ -42,6 +45,7 @@ export function useChat(): ChatState {
       .then((s) => {
         if (alive && s.configured) {
           setLive(true)
+          setModel(s.model)
           setSeed([LIVE_GREETING])
         }
       })
@@ -97,5 +101,5 @@ export function useChat(): ChatState {
     [live],
   )
 
-  return { msgs: [...seed, ...convo], thinking, live, steps, send }
+  return { msgs: [...seed, ...convo], thinking, live, model, steps, send }
 }