chore(model-gallery): ⬆️ update checksum (#10469)

⬆️ Checksum updates in gallery/index.yaml

Signed-off-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: mudler <2420543+mudler@users.noreply.github.com>

backend/cpp/llama-cpp/Makefile

@@ -1,5 +1,5 @@
 
-LLAMA_VERSION?=7c082bc417bbe53210a83df4ba5b49e18ce6193c
+LLAMA_VERSION?=73618f27a801c0b8614ceaf3547d3c2a99baae14
 LLAMA_REPO?=https://github.com/ggerganov/llama.cpp
 
 CMAKE_ARGS?=

backend/go/crispasr/Makefile

@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)
 
 # CrispASR version (release tag)
 CRISPASR_REPO?=https://github.com/CrispStrobe/CrispASR
-CRISPASR_VERSION?=7a8cb80907341c0204bd0488c1244764f4163883
+CRISPASR_VERSION?=63b57289255267edf66e43e33bc3911e04a2e92d
 SO_TARGET?=libgocrispasr.so
 
 CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF

backend/go/stablediffusion-ggml/Makefile

@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)
 
 # stablediffusion.cpp (ggml)
 STABLEDIFFUSION_GGML_REPO?=https://github.com/leejet/stable-diffusion.cpp
-STABLEDIFFUSION_GGML_VERSION?=b12098f5d09fc83da36e65c784f7bdb16a5a5ebf
+STABLEDIFFUSION_GGML_VERSION?=f440ad9c29dd8bc34e5d1f4b863832b96d6ea05f
 
 CMAKE_ARGS+=-DGGML_MAX_NAME=128
 

backend/go/whisper/Makefile

@@ -8,7 +8,7 @@ JOBS?=$(shell nproc --ignore=1)
 
 # whisper.cpp version
 WHISPER_REPO?=https://github.com/ggml-org/whisper.cpp
-WHISPER_CPP_VERSION?=5ed76e9a079962f1c85cfce44edd325c27ef1f97
+WHISPER_CPP_VERSION?=bae6bc02b1940bbfb87b6a0299c565e563b916d1
 SO_TARGET?=libgowhisper.so
 
 CMAKE_ARGS+=-DBUILD_SHARED_LIBS=OFF

core/config/runtime_settings_persist.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"os"
 	"path/filepath"
+	"reflect"
 )
 
 // runtimeSettingsFile is the on-disk filename inside DynamicConfigsDir.
@@ -33,6 +34,35 @@ func (o *ApplicationConfig) ReadPersistedSettings() (RuntimeSettings, error) {
 	return settings, nil
 }
 
+// MergeNonNil overlays every set (non-nil) field of overlay onto the
+// receiver, leaving the receiver's value untouched wherever overlay left a
+// field unset. Every RuntimeSettings field is a pointer precisely so "set"
+// can be told apart from "absent" (see the type doc), which makes this a
+// faithful partial update: a caller that submits only the field it owns
+// changes exactly that field and never clobbers unrelated settings.
+//
+// This is the read-modify-write contract the persistence helpers exist for.
+// UpdateSettingsEndpoint reads the on-disk settings, merges the request body
+// on top, and writes the result — so a focused admin page that POSTs only its
+// own field (the Middleware page sends only mitm_listen; the detector table
+// only pii_default_detectors) no longer nulls every other setting.
+//
+// Reflection keeps the merge total over the struct: a field added to
+// RuntimeSettings later is merged automatically, so the persistence path can
+// never silently drop a new setting the way a hand-maintained field list
+// would. Non-pointer fields (none today) are skipped — they cannot express
+// "absent", so the receiver wins.
+func (s *RuntimeSettings) MergeNonNil(overlay RuntimeSettings) {
+	dst := reflect.ValueOf(s).Elem()
+	src := reflect.ValueOf(overlay)
+	for i := 0; i < src.NumField(); i++ {
+		f := src.Field(i)
+		if f.Kind() == reflect.Pointer && !f.IsNil() {
+			dst.Field(i).Set(f)
+		}
+	}
+}
+
 // WritePersistedSettings serialises the given RuntimeSettings to
 // runtime_settings.json with restricted permissions (it may carry API
 // keys and P2P tokens).

core/config/runtime_settings_persist_test.go

@@ -12,6 +12,7 @@ import (
 )
 
 func strPtr(s string) *string { return &s }
+func boolPtr(b bool) *bool     { return &b }
 
 var _ = Describe("RuntimeSettings persistence helpers", func() {
 	var (
@@ -51,6 +52,47 @@ var _ = Describe("RuntimeSettings persistence helpers", func() {
 		})
 	})
 
+	// MergeNonNil is the partial-update primitive UpdateSettingsEndpoint
+	// relies on: a focused admin page POSTs only the field it owns, and the
+	// handler reads the on-disk settings and overlays the request on top.
+	// Without it, the body would be written verbatim and every field the
+	// caller omitted would be nulled (the reported regression: changing
+	// mitm_listen wiped the galleries, api keys, watchdog config, etc.).
+	Describe("MergeNonNil partial update", func() {
+		It("overlays set fields and preserves unset ones", func() {
+			base := config.RuntimeSettings{
+				MITMListen:          strPtr(":9000"),
+				Galleries:           &[]config.Gallery{{Name: "g1", URL: "http://example/g1"}},
+				WatchdogIdleEnabled: boolPtr(true),
+				ApiKeys:             &[]string{"persisted-key"},
+				PIIDefaultDetectors: &[]string{"det-a"},
+			}
+
+			// Simulate the Middleware proxy tab: only mitm_listen is sent.
+			overlay := config.RuntimeSettings{MITMListen: strPtr(":8443")}
+			base.MergeNonNil(overlay)
+
+			Expect(base.MITMListen).ToNot(BeNil())
+			Expect(*base.MITMListen).To(Equal(":8443"), "set field should be overlaid")
+			// Everything the overlay left unset must survive untouched.
+			Expect(base.Galleries).ToNot(BeNil(), "galleries were clobbered")
+			Expect(*base.Galleries).To(HaveLen(1))
+			Expect(base.WatchdogIdleEnabled).ToNot(BeNil())
+			Expect(*base.WatchdogIdleEnabled).To(BeTrue())
+			Expect(base.ApiKeys).ToNot(BeNil(), "api_keys were clobbered")
+			Expect(*base.ApiKeys).To(Equal([]string{"persisted-key"}))
+			Expect(base.PIIDefaultDetectors).ToNot(BeNil(), "pii_default_detectors were clobbered")
+			Expect(*base.PIIDefaultDetectors).To(Equal([]string{"det-a"}))
+		})
+
+		It("lets an explicit empty slice clear a field", func() {
+			base := config.RuntimeSettings{PIIDefaultDetectors: &[]string{"det-a"}}
+			base.MergeNonNil(config.RuntimeSettings{PIIDefaultDetectors: &[]string{}})
+			Expect(base.PIIDefaultDetectors).ToNot(BeNil())
+			Expect(*base.PIIDefaultDetectors).To(BeEmpty(), "an explicit empty slice should clear, not preserve")
+		})
+	})
+
 	// MITM round trip pins the contract that loadRuntimeSettingsFromFile
 	// MITM listener address must survive a write/read round trip so the
 	// next process restart can bring the listener back up. (Intercept

core/http/endpoints/localai/settings.go

@@ -4,8 +4,6 @@ import (
 	"encoding/json"
 	"io"
 	"net/http"
-	"os"
-	"path/filepath"
 	"time"
 
 	"github.com/labstack/echo/v4"
@@ -110,6 +108,18 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 			})
 		}
 
+		// Read whatever is already persisted: it is both the source of truth
+		// for branding asset filenames (below) and the base we merge this
+		// request onto before writing. A read failure must not let a Save
+		// silently discard the existing settings — surface it instead.
+		persisted, err := appConfig.ReadPersistedSettings()
+		if err != nil {
+			return c.JSON(http.StatusInternalServerError, schema.SettingsResponse{
+				Success: false,
+				Error:   "Failed to read existing settings: " + err.Error(),
+			})
+		}
+
 		// Branding asset filenames are owned exclusively by
 		// /api/branding/asset/{kind} (upload/delete). The Settings page also
 		// round-trips them via GET /api/settings, but its local state is stale
@@ -118,11 +128,9 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 		// at page open. Replace whatever the body sent for these three fields
 		// with the values currently on disk so /api/settings can never
 		// regress them.
-		if existing, err := appConfig.ReadPersistedSettings(); err == nil {
-			settings.LogoFile = existing.LogoFile
-			settings.LogoHorizontalFile = existing.LogoHorizontalFile
-			settings.FaviconFile = existing.FaviconFile
-		}
+		settings.LogoFile = persisted.LogoFile
+		settings.LogoHorizontalFile = persisted.LogoHorizontalFile
+		settings.FaviconFile = persisted.FaviconFile
 
 		// The UI reads ApiKeys from GET /api/settings, which already returns the
 		// merged env+runtime list. When the user clicks Save, the same merged
@@ -145,16 +153,17 @@ func UpdateSettingsEndpoint(app *application.Application) echo.HandlerFunc {
 			settings.ApiKeys = &runtimeOnly
 		}
 
-		settingsFile := filepath.Join(appConfig.DynamicConfigsDir, "runtime_settings.json")
-		settingsJSON, err := json.MarshalIndent(settings, "", "  ")
-		if err != nil {
-			return c.JSON(http.StatusInternalServerError, schema.SettingsResponse{
-				Success: false,
-				Error:   "Failed to marshal settings: " + err.Error(),
-			})
-		}
-
-		if err := os.WriteFile(settingsFile, settingsJSON, 0600); err != nil {
+		// Persist as a partial update: overlay only the fields this request set
+		// onto the settings already on disk. Focused admin pages POST just the
+		// keys they own (the Middleware proxy tab sends only mitm_listen; the
+		// detector table only pii_default_detectors), so writing the request
+		// body verbatim would null every unrelated setting (the no-omitempty
+		// api_keys / pii_default_detectors fields even round-trip as JSON
+		// null). The full Settings page still round-trips every field, so its
+		// Save is unchanged.
+		toPersist := persisted
+		toPersist.MergeNonNil(settings)
+		if err := appConfig.WritePersistedSettings(toPersist); err != nil {
 			return c.JSON(http.StatusInternalServerError, schema.SettingsResponse{
 				Success: false,
 				Error:   "Failed to write settings file: " + err.Error(),

core/http/endpoints/localai/settings_test.go

@@ -52,6 +52,10 @@ var _ = Describe("Settings endpoints", func() {
 		// Settings are persisted here; set after construction since there's no
 		// dedicated AppOption for it.
 		app.ApplicationConfig().DynamicConfigsDir = tmp
+		// Contain the MITM CA inside tmp too. The partial-save spec flips
+		// mitm_listen, which starts the listener and writes a CA; without this
+		// it defaults to ./mitm-ca and litters the package source tree.
+		app.ApplicationConfig().MITMCADir = filepath.Join(tmp, "mitm-ca")
 
 		e = echo.New()
 		e.GET("/api/settings", GetSettingsEndpoint(app))
@@ -109,6 +113,39 @@ var _ = Describe("Settings endpoints", func() {
 		Expect(err).ToNot(HaveOccurred())
 	})
 
+	// Regression: a focused admin page (the Middleware proxy tab) POSTs only
+	// the one field it owns — mitm_listen. The old handler wrote the request
+	// body verbatim, so every other persisted setting was dropped (and
+	// api_keys / pii_default_detectors, which lack omitempty, were written as
+	// null). A partial POST must now merge onto what is already on disk.
+	It("preserves unrelated persisted settings when a partial POST sets only mitm_listen", func() {
+		// First save establishes a fuller settings file (as the full Settings
+		// page would): galleries, an API key, and the MITM listener. The
+		// listener restart binds a real socket, so use 127.0.0.1:0 for an
+		// ephemeral free port rather than a fixed one that may be in use.
+		rec := post(`{"mitm_listen":"127.0.0.1:0","galleries":[{"name":"g1","url":"http://example/g1"}],"api_keys":["k1"],"pii_default_detectors":["det-a"]}`)
+		Expect(rec.Code).To(Equal(http.StatusOK), rec.Body.String())
+
+		// The Middleware proxy tab then changes only the listen address — the
+		// exact partial body that nulled everything else before the fix.
+		rec = post(`{"mitm_listen":"127.0.0.1:0"}`)
+		Expect(rec.Code).To(Equal(http.StatusOK), rec.Body.String())
+
+		raw, err := os.ReadFile(filepath.Join(tmp, "runtime_settings.json"))
+		Expect(err).ToNot(HaveOccurred())
+		var ondisk config.RuntimeSettings
+		Expect(json.Unmarshal(raw, &ondisk)).To(Succeed())
+
+		Expect(ondisk.MITMListen).ToNot(BeNil())
+		Expect(*ondisk.MITMListen).To(Equal("127.0.0.1:0"), "the changed field should be saved")
+		Expect(ondisk.Galleries).ToNot(BeNil(), "galleries were clobbered by the partial save")
+		Expect(*ondisk.Galleries).To(HaveLen(1))
+		Expect(ondisk.ApiKeys).ToNot(BeNil(), "api_keys were nulled by the partial save")
+		Expect(*ondisk.ApiKeys).To(Equal([]string{"k1"}))
+		Expect(ondisk.PIIDefaultDetectors).ToNot(BeNil(), "pii_default_detectors were nulled by the partial save")
+		Expect(*ondisk.PIIDefaultDetectors).To(Equal([]string{"det-a"}))
+	})
+
 	// Residual #9125: enabling the watchdog from a cold (off) state via the
 	// React master toggle must start the live watchdog immediately, without a
 	// restart. The toggle posts watchdog_idle_enabled/busy_enabled=true while

core/http/react-ui/e2e/role-mode-adaptive.spec.js

@@ -1,100 +0,0 @@
-import { test, expect } from './coverage-fixtures.js'
-
-// These specs stub /api/features and /api/auth/status per cell. The test server
-// disables auth (isAdmin=true) and reports its own features, so we intercept
-// before navigation to simulate each role x mode cell.
-
-function stubFeatures(page, features) {
-  return page.route('**/api/features', route =>
-    route.fulfill({ contentType: 'application/json', body: JSON.stringify(features) }))
-}
-
-function stubNoP2P(page) {
-  // P2P token endpoint returns empty -> p2pEnabled=false.
-  return page.route('**/api/p2p/token', route =>
-    route.fulfill({ contentType: 'text/plain', body: '' }))
-}
-
-test.describe('Adaptive landing (HomeRoute)', () => {
-  test('admin + distributed redirects /app to Nodes', async ({ page }) => {
-    await stubFeatures(page, { distributed: true })
-    await stubNoP2P(page)
-    await page.goto('/app')
-    await expect(page).toHaveURL(/\/app\/nodes$/)
-    await expect(page.locator('.page-title').first()).toBeVisible({ timeout: 15_000 })
-  })
-
-  test('admin + single-node stays on Home', async ({ page }) => {
-    await stubFeatures(page, { distributed: false })
-    await stubNoP2P(page)
-    await page.goto('/app')
-    await expect(page).toHaveURL(/\/app$/)
-    await expect(page.locator('.home-greeting')).toBeVisible({ timeout: 15_000 })
-  })
-})
-
-test.describe('Adaptive sidebar', () => {
-  test('distributed pins the Cluster group with Nodes at the top', async ({ page }) => {
-    await stubFeatures(page, { distributed: true })
-    await stubNoP2P(page)
-    await page.goto('/app/chat') // any in-app page so the sidebar is mounted
-    const pinned = page.locator('.sidebar-nav .sidebar-section-items').first()
-    await expect(pinned.getByText('Nodes', { exact: false })).toBeVisible({ timeout: 15_000 })
-  })
-
-  test('single-node does not pin a Cluster group', async ({ page }) => {
-    await stubFeatures(page, { distributed: false })
-    await stubNoP2P(page)
-    await page.goto('/app/chat')
-    // Nodes is reachable only via the Operate rail, not pinned at the top.
-    await expect(page.locator('.sidebar-nav')).toBeVisible({ timeout: 15_000 })
-    await expect(page.locator('.sidebar-nav .sidebar-section-items').first()
-      .getByText('Nodes', { exact: false })).toHaveCount(0)
-  })
-})
-
-test.describe('Top navbar', () => {
-  test('admin sees the mode pill and settings cog', async ({ page }) => {
-    await stubFeatures(page, { distributed: true })
-    await stubNoP2P(page)
-    await page.goto('/app/chat')
-    await expect(page.locator('.top-navbar__mode')).toBeVisible({ timeout: 15_000 })
-    await expect(page.locator('.top-navbar__icon[aria-label]')).not.toHaveCount(0)
-  })
-
-  test('admin-via-chat jump shows when localai_assistant is enabled', async ({ page }) => {
-    await stubFeatures(page, { distributed: false, localai_assistant: true })
-    await stubNoP2P(page)
-    await page.goto('/app/chat')
-    await expect(page.locator('.top-navbar__assistant')).toBeVisible({ timeout: 15_000 })
-  })
-
-  test('admin-via-chat jump hidden when localai_assistant is off', async ({ page }) => {
-    await stubFeatures(page, { distributed: false, localai_assistant: false })
-    await stubNoP2P(page)
-    await page.goto('/app/chat')
-    await expect(page.locator('.top-navbar__assistant')).toHaveCount(0)
-  })
-})
-
-test.describe('Token usage meter', () => {
-  test('renders when admin usage has data', async ({ page }) => {
-    await stubFeatures(page, { distributed: false })
-    await stubNoP2P(page)
-    await page.route('**/api/auth/admin/usage**', route =>
-      route.fulfill({ contentType: 'application/json',
-        body: JSON.stringify({ buckets: [{ total_tokens: 1234 }] }) }))
-    await page.goto('/app/chat')
-    await expect(page.locator('.top-navbar__meter')).toBeVisible({ timeout: 15_000 })
-  })
-
-  test('hidden when admin usage is empty (graceful degrade)', async ({ page }) => {
-    await stubFeatures(page, { distributed: false })
-    await stubNoP2P(page)
-    await page.route('**/api/auth/admin/usage**', route =>
-      route.fulfill({ contentType: 'application/json', body: JSON.stringify({ buckets: [] }) }))
-    await page.goto('/app/chat')
-    await expect(page.locator('.top-navbar')).toBeVisible({ timeout: 15_000 })
-    await expect(page.locator('.top-navbar__meter')).toHaveCount(0)
-  })
-})

core/http/react-ui/public/locales/en/nav.json

@@ -12,16 +12,6 @@
   "accountSettings": "Account settings",
   "account": "Account",
   "accountFor": "Account: {{name}}",
-  "topbar": {
-    "label": "Top bar",
-    "modeDistributed": "Distributed",
-    "modeSwarm": "Swarm",
-    "modeSingle": "Single-node",
-    "pickModel": "Models",
-    "adminViaChat": "Admin via chat",
-    "tokensToday": "Tokens today",
-    "usageDetail": "View usage detail"
-  },
   "sections": {
     "create": "Create",
     "recognition": "Recognition",

core/http/react-ui/src/App.css

@@ -184,50 +184,6 @@
   font-size: 1.5rem;
 }
 
-/* Desktop top bar: deployment + admin affordances on wide screens. Hidden on
-   mobile, where .mobile-header carries the equivalent actions. */
-.top-navbar {
-  display: flex;
-  align-items: center;
-  justify-content: space-between;
-  gap: var(--spacing-md);
-  padding: var(--spacing-sm) var(--spacing-lg);
-  border-bottom: 1px solid var(--color-border-default);
-  background: var(--color-bg-secondary);
-}
-.top-navbar__right { display: flex; align-items: center; gap: var(--spacing-sm); }
-.top-navbar__mode {
-  font-size: 0.75rem;
-  padding: 2px 10px;
-  border-radius: 999px;
-  border: 1px solid var(--color-border-default);
-  color: var(--color-text-secondary);
-}
-.top-navbar__mode.is-active { color: var(--color-success); border-color: var(--color-success); }
-.top-navbar__btn {
-  display: inline-flex; align-items: center; gap: 6px;
-  font-size: 0.8125rem; padding: 5px 10px; border-radius: 8px;
-  border: 1px solid var(--color-border-default); background: var(--color-bg-tertiary);
-  color: var(--color-text-primary); cursor: pointer;
-}
-.top-navbar__icon {
-  width: 32px; height: 32px; display: inline-flex; align-items: center;
-  justify-content: center; border-radius: 8px; border: 1px solid var(--color-border-default);
-  background: var(--color-bg-tertiary); color: var(--color-text-secondary); cursor: pointer;
-}
-.top-navbar__avatar img { width: 100%; height: 100%; border-radius: 50%; object-fit: cover; }
-.top-navbar__meter {
-  display: inline-flex; flex-direction: column; gap: 3px; align-items: flex-start;
-  padding: 4px 10px; border-radius: 8px; border: 1px solid var(--color-border-default);
-  background: var(--color-bg-tertiary); cursor: pointer; min-width: 150px;
-}
-.top-navbar__meter-label { font-size: 0.6875rem; color: var(--color-text-secondary); }
-.top-navbar__meter-bar { width: 100%; height: 5px; border-radius: 3px; background: var(--color-bg-secondary); overflow: hidden; }
-.top-navbar__meter-bar i { display: block; height: 100%; background: var(--color-primary); }
-@media (max-width: 639px) {
-  .top-navbar { display: none; }
-}
-
 /* Sidebar */
 .sidebar {
   position: fixed;

core/http/react-ui/src/App.jsx

@@ -3,7 +3,6 @@ import { Outlet, useLocation, useNavigate } from 'react-router-dom'
 import { useTranslation } from 'react-i18next'
 import Sidebar from './components/Sidebar'
 import OperationsBar from './components/OperationsBar'
-import TopNavbar from './components/TopNavbar'
 import { ToastContainer, useToast } from './components/Toast'
 import { systemApi } from './utils/api'
 import { useTheme } from './contexts/ThemeContext'
@@ -99,7 +98,6 @@ export default function App() {
       <Sidebar isOpen={sidebarOpen} onClose={() => setSidebarOpen(false)} />
       <main className="main-content" {...(sidebarOpen ? { 'aria-hidden': 'true', inert: '' } : {})}>
         <OperationsBar />
-        <TopNavbar />
         {/* Mobile header — primary actions reachable without opening the
             drawer. Hamburger is the only way to expand the nav on phones;
             theme toggle and account avatar are mirrored from the sidebar

core/http/react-ui/src/components/HomeRoute.jsx

@@ -1,28 +0,0 @@
-import { lazy, Suspense } from 'react'
-import { Navigate } from 'react-router-dom'
-import { useAuth } from '../context/AuthContext'
-import { useDeployment } from '../contexts/DeploymentContext'
-import { resolveHome } from '../utils/resolveHome'
-import RouteFallback from './RouteFallback'
-
-const Home = lazy(() => import('../pages/Home'))
-
-// Index-route element. Waits for auth + deployment signals to load (so we never
-// flash the wrong landing), then either renders Home or redirects to the cell's
-// landing page. Redirecting (rather than rendering Nodes/Chat inline at /app)
-// keeps each target's own route guard, active-nav state, and deep-linkability.
-export default function HomeRoute() {
-  const { isAdmin, loading: authLoading } = useAuth()
-  const { distributed, p2pEnabled, loading: deployLoading } = useDeployment()
-
-  if (authLoading || deployLoading) return <RouteFallback />
-
-  const target = resolveHome({ isAdmin, distributed, p2pEnabled })
-  if (target) return <Navigate to={target} replace />
-
-  return (
-    <Suspense fallback={<RouteFallback />}>
-      <Home />
-    </Suspense>
-  )
-}

core/http/react-ui/src/components/Sidebar.jsx

@@ -5,11 +5,9 @@ import ThemeToggle from './ThemeToggle'
 import LanguageSwitcher from './LanguageSwitcher'
 import { useAuth } from '../context/AuthContext'
 import { useBranding } from '../contexts/BrandingContext'
-import { useDeployment } from '../contexts/DeploymentContext'
 import { apiUrl } from '../utils/basePath'
 import { preloadRoute } from '../router'
 import { consoles, firstVisiblePath, consolePaths } from './console/consoleConfig'
-import { clusterPinItems, shouldCollapseCreate } from '../utils/sidebarPolicy'
 
 const COLLAPSED_KEY = 'localai_sidebar_collapsed'
 const SECTIONS_KEY = 'localai_sidebar_sections'
@@ -60,13 +58,11 @@ function NavItem({ item, onClose, collapsed }) {
   )
 }
 
-function loadSectionState(collapseCreate = false) {
-  // Tiers render expanded by default; users can collapse any tier and the
-  // choice persists (stored values override defaults). In cluster cells we
-  // start Create collapsed so the pinned cluster group leads - but only when
-  // the user has not already expressed a preference.
+function loadSectionState() {
+  // Tiers render expanded by default (the redesign favours showing the few
+  // intent groups up front); users can still collapse any tier and the choice
+  // is persisted. Stored values override the defaults so a saved collapse wins.
   const defaults = Object.fromEntries(sections.map(s => [s.id, true]))
-  if (collapseCreate) defaults.create = false
   try {
     const stored = localStorage.getItem(SECTIONS_KEY)
     return stored ? { ...defaults, ...JSON.parse(stored) } : defaults
@@ -81,34 +77,20 @@ function saveSectionState(state) {
 
 export default function Sidebar({ isOpen, onClose }) {
   const { t } = useTranslation('nav')
-  const { isAdmin, authEnabled, user, logout, hasFeature } = useAuth()
-  // Deployment shape (server features + p2p) drives the adaptive sidebar; the
-  // shared context replaces the sidebar's own /api/features fetch so the
-  // landing resolver, navbar, and this policy agree on one snapshot.
-  const deployment = useDeployment()
-  const features = deployment.features
-  // Shared shape for the console gating helpers (consoleConfig.js); in scope for
-  // both the pinned cluster group and the console-tier rendering below.
-  const auth = { isAdmin, authEnabled, hasFeature, features }
-  const collapseCreate = shouldCollapseCreate(auth, deployment)
+  const [features, setFeatures] = useState({})
   const [collapsed, setCollapsed] = useState(() => {
     try { return localStorage.getItem(COLLAPSED_KEY) === 'true' } catch (_) { return false }
   })
   const [openSections, setOpenSections] = useState(loadSectionState)
+  const { isAdmin, authEnabled, user, logout, hasFeature } = useAuth()
   const branding = useBranding()
   const navigate = useNavigate()
   const location = useLocation()
   const closeBtnRef = useRef(null)
 
-  // Apply the cluster-cell Create-collapse default once, only when the user has
-  // no stored section preference (so we never override an explicit choice).
   useEffect(() => {
-    if (deployment.loading) return
-    let hasStored = false
-    try { hasStored = !!localStorage.getItem(SECTIONS_KEY) } catch { hasStored = false }
-    if (hasStored || !collapseCreate) return
-    setOpenSections(prev => (prev.create === false ? prev : { ...prev, create: false }))
-  }, [deployment.loading, collapseCreate])
+    fetch(apiUrl('/api/features')).then(r => r.json()).then(setFeatures).catch(() => {})
+  }, [])
 
   // Stay in sync with external collapse dispatches (e.g. the chat
   // page's focus mode). The collapse-toggle button still owns the
@@ -175,6 +157,8 @@ export default function Sidebar({ isOpen, onClose }) {
   }
 
   const visibleTopItems = topItems.filter(filterItem)
+  // Shared shape for the console gating helpers (consoleConfig.js).
+  const auth = { isAdmin, authEnabled, hasFeature, features }
 
   // Inline sections (Create) carry no gating; a plain filterItem pass suffices.
   const getVisibleSectionItems = (section) => section.items.filter(filterItem)
@@ -215,28 +199,6 @@ export default function Sidebar({ isOpen, onClose }) {
             ))}
           </div>
 
-          {/* Pinned Cluster quick-access (admin + distributed/p2p). Same gate
-              as the Operate rail; surfaced at the top for cluster operators. */}
-          {(() => {
-            const pinned = clusterPinItems(auth, deployment)
-            if (pinned.length === 0) return null
-            return (
-              <div className="sidebar-section">
-                <div className="sidebar-section-title">{t('operate.cluster')}</div>
-                <div className="sidebar-section-items">
-                  {pinned.map(item => (
-                    <NavItem
-                      key={item.path}
-                      item={{ path: item.path, icon: item.icon, labelKey: item.labelKey }}
-                      onClose={onClose}
-                      collapsed={collapsed}
-                    />
-                  ))}
-                </div>
-              </div>
-            )
-          })()}
-
           {/* Collapsible sections */}
           {sections.map(section => {
             const visibleItems = getVisibleSectionItems(section)

core/http/react-ui/src/components/TopNavbar.jsx

@@ -1,96 +0,0 @@
-import { useNavigate } from 'react-router-dom'
-import { useTranslation } from 'react-i18next'
-import { useAuth } from '../context/AuthContext'
-import { useDeployment } from '../contexts/DeploymentContext'
-import { useTheme } from '../contexts/ThemeContext'
-import { launchAssistantChat } from '../utils/launchAssistantChat'
-import TokenUsageMeter from './navbar/TokenUsageMeter'
-
-// Desktop top bar. Complementary to the mobile-only header in App.jsx: this is
-// hidden on small screens (see .top-navbar CSS) and shows deployment/admin
-// affordances on wide screens where the sidebar footer is far from the content.
-export default function TopNavbar() {
-  const { t } = useTranslation('nav')
-  const navigate = useNavigate()
-  const { isAdmin, authEnabled, user } = useAuth()
-  const { features, distributed, p2pEnabled } = useDeployment()
-  const { theme, toggleTheme } = useTheme()
-
-  const modeLabel = distributed
-    ? t('topbar.modeDistributed')
-    : p2pEnabled
-      ? t('topbar.modeSwarm')
-      : t('topbar.modeSingle')
-
-  const showAssistantJump = isAdmin && !!features.localai_assistant
-  const showAvatar = authEnabled && user
-  const themeLabel = theme === 'dark' ? t('switchToLightMode') : t('switchToDarkMode')
-
-  return (
-    <div className="top-navbar" role="navigation" aria-label={t('topbar.label')}>
-      <div className="top-navbar__left">
-        {isAdmin && (
-          <span className={`top-navbar__mode ${distributed || p2pEnabled ? 'is-active' : ''}`}>
-            <i className="fas fa-circle-nodes" aria-hidden="true" /> {modeLabel}
-          </span>
-        )}
-      </div>
-      <div className="top-navbar__right">
-        {!isAdmin && (
-          <button
-            type="button"
-            className="top-navbar__btn"
-            onClick={() => navigate('/app/chat')}
-            title={t('topbar.pickModel')}
-          >
-            <i className="fas fa-cube" aria-hidden="true" /> {t('topbar.pickModel')}
-          </button>
-        )}
-        {showAssistantJump && (
-          <button
-            type="button"
-            className="top-navbar__btn top-navbar__assistant"
-            onClick={() => launchAssistantChat(navigate)}
-            title={t('topbar.adminViaChat')}
-          >
-            <i className="fas fa-user-shield" aria-hidden="true" /> {t('topbar.adminViaChat')}
-          </button>
-        )}
-        {isAdmin && <TokenUsageMeter />}
-        {isAdmin && (
-          <button
-            type="button"
-            className="top-navbar__icon"
-            onClick={() => navigate('/app/settings')}
-            aria-label={t('items.settings')}
-            title={t('items.settings')}
-          >
-            <i className="fas fa-cog" aria-hidden="true" />
-          </button>
-        )}
-        <button
-          type="button"
-          className="top-navbar__icon"
-          onClick={toggleTheme}
-          aria-label={themeLabel}
-          title={themeLabel}
-        >
-          <i className={`fas ${theme === 'dark' ? 'fa-sun' : 'fa-moon'}`} aria-hidden="true" />
-        </button>
-        {showAvatar && (
-          <button
-            type="button"
-            className="top-navbar__icon top-navbar__avatar"
-            onClick={() => navigate('/app/account')}
-            aria-label={user.name || user.email}
-            title={user.name || user.email}
-          >
-            {user.avatarUrl
-              ? <img src={user.avatarUrl} alt="" />
-              : <i className="fas fa-user-circle" aria-hidden="true" />}
-          </button>
-        )}
-      </div>
-    </div>
-  )
-}

core/http/react-ui/src/components/navbar/TokenUsageMeter.jsx

@@ -1,52 +0,0 @@
-import { useState, useEffect } from 'react'
-import { useNavigate } from 'react-router-dom'
-import { useTranslation } from 'react-i18next'
-import { usageApi } from '../../utils/api'
-
-// Compact admin-only usage glance: today's total tokens, optionally against a
-// quota cap, linking to the full /app/usage page. Self-contained data fetch so
-// a usage-API failure cannot break the navbar - it just renders nothing.
-function sumTotalTokens(res) {
-  const buckets = res?.buckets || res?.usage || (Array.isArray(res) ? res : [])
-  if (!Array.isArray(buckets) || buckets.length === 0) return null
-  return buckets.reduce((s, b) => s + (b.total_tokens || 0), 0)
-}
-
-export default function TokenUsageMeter() {
-  const { t } = useTranslation('nav')
-  const navigate = useNavigate()
-  const [tokens, setTokens] = useState(null)
-  const [cap, setCap] = useState(null)
-
-  useEffect(() => {
-    let cancelled = false
-    usageApi.getAdminUsage('day')
-      .then(res => { if (!cancelled) setTokens(sumTotalTokens(res)) })
-      .catch(() => { if (!cancelled) setTokens(null) })
-    usageApi.getMyQuotas()
-      .then(q => { if (!cancelled) setCap(q?.token_limit || q?.tokens?.limit || null) })
-      .catch(() => { if (!cancelled) setCap(null) })
-    return () => { cancelled = true }
-  }, [])
-
-  if (tokens === null) return null
-
-  const pct = cap ? Math.min(100, Math.round((tokens / cap) * 100)) : null
-
-  return (
-    <button
-      type="button"
-      className="top-navbar__meter"
-      onClick={() => navigate('/app/usage')}
-      title={t('topbar.usageDetail')}
-    >
-      <span className="top-navbar__meter-label">
-        {t('topbar.tokensToday')}: {Intl.NumberFormat().format(tokens)}
-        {cap ? ` / ${Intl.NumberFormat().format(cap)}` : ''}
-      </span>
-      {pct !== null && (
-        <span className="top-navbar__meter-bar"><i style={{ width: `${pct}%` }} /></span>
-      )}
-    </button>
-  )
-}

core/http/react-ui/src/contexts/DeploymentContext.jsx

@@ -1,55 +0,0 @@
-import { createContext, useContext, useState, useEffect } from 'react'
-import { apiUrl } from '../utils/basePath'
-import { p2pApi } from '../utils/api'
-
-const DeploymentContext = createContext(null)
-
-// One shared fetch of the deployment-shape signals the adaptive UI keys off:
-// server features (/api/features) and whether a P2P network token exists.
-// Components used to fetch /api/features independently (Sidebar, Home); this
-// centralises it so the landing resolver, sidebar policy, and navbar agree on
-// one snapshot and we issue a single request.
-export function DeploymentProvider({ children }) {
-  const [features, setFeatures] = useState({})
-  const [p2pEnabled, setP2pEnabled] = useState(false)
-  const [loading, setLoading] = useState(true)
-
-  useEffect(() => {
-    let cancelled = false
-    const featuresP = fetch(apiUrl('/api/features'))
-      .then(r => r.json())
-      .catch(() => ({}))
-    // P2P has no /api/features flag: it is "enabled" when a network token
-    // exists (mirrors pages/P2P.jsx). A 404/disabled endpoint throws and we
-    // treat that as not-enabled.
-    const p2pP = p2pApi.getToken()
-      .then(tok => (typeof tok === 'string' ? tok : (tok?.token || '')).trim())
-      .catch(() => '')
-    Promise.all([featuresP, p2pP]).then(([f, tok]) => {
-      if (cancelled) return
-      setFeatures(f || {})
-      setP2pEnabled(!!tok)
-      setLoading(false)
-    })
-    return () => { cancelled = true }
-  }, [])
-
-  const value = {
-    features,
-    distributed: !!features.distributed,
-    p2pEnabled,
-    loading,
-  }
-
-  return (
-    <DeploymentContext.Provider value={value}>
-      {children}
-    </DeploymentContext.Provider>
-  )
-}
-
-export function useDeployment() {
-  const ctx = useContext(DeploymentContext)
-  if (!ctx) throw new Error('useDeployment must be used within DeploymentProvider')
-  return ctx
-}

core/http/react-ui/src/main.jsx

@@ -4,7 +4,6 @@ import { RouterProvider } from 'react-router-dom'
 import { ThemeProvider } from './contexts/ThemeContext'
 import { BrandingProvider } from './contexts/BrandingContext'
 import { AuthProvider } from './context/AuthContext'
-import { DeploymentProvider } from './contexts/DeploymentContext'
 import { OperationsProvider } from './contexts/OperationsContext'
 import { router } from './router'
 import './i18n'
@@ -33,11 +32,9 @@ createRoot(document.getElementById('root')).render(
       <ThemeProvider>
         <BrandingProvider>
           <AuthProvider>
-            <DeploymentProvider>
-              <OperationsProvider>
-                <RouterProvider router={router} />
-              </OperationsProvider>
-            </DeploymentProvider>
+            <OperationsProvider>
+              <RouterProvider router={router} />
+            </OperationsProvider>
           </AuthProvider>
         </BrandingProvider>
       </ThemeProvider>

core/http/react-ui/src/pages/Chat.jsx

@@ -541,73 +541,58 @@ export default function Chat() {
     updateChatSettings(activeChat.id, { clientMCPServers: next })
   }, [activeChat, updateChatSettings])
 
-  // Load initial message / assistant launch from the Home page or the navbar
-  // quick-jump. Factored into a callback so both the mount-time reader and the
-  // navbar re-trigger event below consume the same payload through one path.
+  // Load initial message from home page
   const homeDataProcessed = useRef(false)
-  const consumeHomeChatData = useCallback(() => {
-    const stored = localStorage.getItem('localai_index_chat_data')
-    if (!stored) return
-    try {
-      const data = JSON.parse(stored)
-      localStorage.removeItem('localai_index_chat_data')
-
-      // Two entry shapes from Home:
-      //   - "compose-and-send": data.message present → open new chat,
-      //     prefill the composer, click submit.
-      //   - "open-assistant": no message, just data.localaiAssistant → open
-      //     a fresh chat already in admin mode so the wizard can fire.
-      const hasMessage = !!data.message
-      const wantsAssistant = !!data.localaiAssistant
-
-      if (hasMessage || wantsAssistant) {
-        let targetChat = activeChat
-        if (data.newChat) {
-          targetChat = addChat(data.model || '', '', data.mcpMode || false)
-        } else {
-          if (data.model && activeChat) {
-            updateChatSettings(activeChat.id, { model: data.model })
-          }
-          if (data.mcpMode && activeChat) {
-            updateChatSettings(activeChat.id, { mcpMode: true })
-          }
-        }
-        if (data.mcpServers?.length > 0 && targetChat) {
-          updateChatSettings(targetChat.id, { mcpServers: data.mcpServers })
-        }
-        if (data.clientMCPServers?.length > 0 && targetChat) {
-          updateChatSettings(targetChat.id, { clientMCPServers: data.clientMCPServers })
-        }
-        if (wantsAssistant && targetChat) {
-          updateChatSettings(targetChat.id, { localaiAssistant: true })
-        }
-        if (hasMessage) {
-          setInput(data.message)
-          if (data.files) setFiles(data.files)
-          setTimeout(() => {
-            const submitBtn = document.getElementById('chat-submit-btn')
-            submitBtn?.click()
-          }, 100)
-        }
-      }
-    } catch (_e) { /* ignore */ }
-  }, [activeChat, addChat, updateChatSettings])
-
   useEffect(() => {
     if (homeDataProcessed.current) return
-    homeDataProcessed.current = true
-    consumeHomeChatData()
-  }, [consumeHomeChatData])
+    const stored = localStorage.getItem('localai_index_chat_data')
+    if (stored) {
+      homeDataProcessed.current = true
+      try {
+        const data = JSON.parse(stored)
+        localStorage.removeItem('localai_index_chat_data')
 
-  // Admins can re-trigger the assistant jump from the navbar while already on
-  // the chat page; navigate('/app/chat') does not remount Chat, so the
-  // mount-time reader above never fires. The launcher dispatches this event
-  // after writing the payload so we re-consume it and open a fresh assistant.
-  useEffect(() => {
-    const onOpenAssistant = () => consumeHomeChatData()
-    window.addEventListener('localai-open-assistant', onOpenAssistant)
-    return () => window.removeEventListener('localai-open-assistant', onOpenAssistant)
-  }, [consumeHomeChatData])
+        // Two entry shapes from Home:
+        //   - "compose-and-send": data.message present → open new chat,
+        //     prefill the composer, click submit.
+        //   - "open-assistant": no message, just data.localaiAssistant → open
+        //     a fresh chat already in admin mode so the wizard can fire.
+        const hasMessage = !!data.message
+        const wantsAssistant = !!data.localaiAssistant
+
+        if (hasMessage || wantsAssistant) {
+          let targetChat = activeChat
+          if (data.newChat) {
+            targetChat = addChat(data.model || '', '', data.mcpMode || false)
+          } else {
+            if (data.model && activeChat) {
+              updateChatSettings(activeChat.id, { model: data.model })
+            }
+            if (data.mcpMode && activeChat) {
+              updateChatSettings(activeChat.id, { mcpMode: true })
+            }
+          }
+          if (data.mcpServers?.length > 0 && targetChat) {
+            updateChatSettings(targetChat.id, { mcpServers: data.mcpServers })
+          }
+          if (data.clientMCPServers?.length > 0 && targetChat) {
+            updateChatSettings(targetChat.id, { clientMCPServers: data.clientMCPServers })
+          }
+          if (wantsAssistant && targetChat) {
+            updateChatSettings(targetChat.id, { localaiAssistant: true })
+          }
+          if (hasMessage) {
+            setInput(data.message)
+            if (data.files) setFiles(data.files)
+            setTimeout(() => {
+              const submitBtn = document.getElementById('chat-submit-btn')
+              submitBtn?.click()
+            }, 100)
+          }
+        }
+      } catch (_e) { /* ignore */ }
+    }
+  }, [])
 
   // Track whether the user is pinned to the bottom. If they scroll up
   // while a response is streaming, stop forcing them back down.

core/http/react-ui/src/pages/Home.jsx

@@ -13,7 +13,6 @@ import { useResources } from '../hooks/useResources'
 import { fileToBase64, backendControlApi, systemApi, modelsApi, mcpApi, nodesApi } from '../utils/api'
 import { API_CONFIG } from '../utils/config'
 import { greetingKey } from '../utils/greeting'
-import { launchAssistantChat } from '../utils/launchAssistantChat'
 import StatusPill from '../components/StatusPill'
 import Skeleton from '../components/Skeleton'
 import SectionHeading from '../components/SectionHeading'
@@ -229,8 +228,16 @@ export default function Home() {
   // requiring an initial message or model selection. Useful when an admin
   // wants to start the assistant from a cold home page.
   const openAssistantChat = useCallback(() => {
-    launchAssistantChat(navigate, selectedModel)
+    const chatData = {
+      model: selectedModel || '',
+      mcpMode: false,
+      localaiAssistant: true,
+      newChat: true,
+    }
+    localStorage.setItem('localai_index_chat_data', JSON.stringify(chatData))
+    try { localStorage.setItem('localai_assistant_used', '1') } catch { /* ignore */ }
     setAssistantUsed(true)
+    navigate('/app/chat')
   }, [navigate, selectedModel])
 
   const handleSubmit = (e) => {

core/http/react-ui/src/router.jsx

@@ -6,7 +6,6 @@ import RequireAdmin from './components/RequireAdmin'
 import RequireAuth from './components/RequireAuth'
 import RequireAuthEnabled from './components/RequireAuthEnabled'
 import RequireFeature from './components/RequireFeature'
-import HomeRoute from './components/HomeRoute'
 
 // Pages are code-split: each becomes its own chunk loaded on demand, so a route
 // no longer drags every other page (and its heavy deps — CodeMirror, the MCP
@@ -33,7 +32,7 @@ export function preloadRoute(path) {
   preloaders[m[1] ?? '']?.().catch(() => { /* network blip — real click will retry */ })
 }
 
-page('', () => import('./pages/Home'))
+const Home = page('', () => import('./pages/Home'))
 const Chat = page('chat', () => import('./pages/Chat'))
 const Models = page('models', () => import('./pages/Models'))
 const Manage = page('manage', () => import('./pages/Manage'))
@@ -97,7 +96,7 @@ function Feature({ feature, children }) {
 }
 
 const appChildren = [
-  { index: true, element: <HomeRoute /> },
+  { index: true, element: <Home /> },
   { path: 'chat', element: <Chat /> },
   { path: 'chat/:model', element: <Chat /> },
   { path: 'image', element: <ImageGen /> },

core/http/react-ui/src/utils/launchAssistantChat.js

@@ -1,19 +0,0 @@
-// Opens a fresh chat already in LocalAI Assistant ("manage") mode. Chat.jsx
-// reads localai_index_chat_data on mount and enables localaiAssistant for the
-// new chat. Shared by the Home CTA and the top navbar quick-jump so there is
-// one definition of how the assistant is launched.
-export function launchAssistantChat(navigate, model = '') {
-  const chatData = {
-    model: model || '',
-    mcpMode: false,
-    localaiAssistant: true,
-    newChat: true,
-  }
-  try { localStorage.setItem('localai_index_chat_data', JSON.stringify(chatData)) } catch { /* ignore */ }
-  try { localStorage.setItem('localai_assistant_used', '1') } catch { /* ignore */ }
-  navigate('/app/chat')
-  // When already on /app/chat, navigate() does not remount Chat, so its
-  // mount-time reader would never see the payload above. Signal the mounted
-  // Chat to re-consume it; harmless elsewhere since Chat reads on mount anyway.
-  try { window.dispatchEvent(new CustomEvent('localai-open-assistant')) } catch { /* ignore */ }
-}

core/http/react-ui/src/utils/resolveHome.js

@@ -1,11 +0,0 @@
-// Pure landing-page resolver for the index route. Returns a target path, or ''
-// meaning "render the default Home". Admin precedence is distributed > p2p >
-// plain; non-admins always go to Chat (distributed/p2p are admin-only and
-// invisible to them). Visibility gates are enforced elsewhere - this only
-// chooses where /app lands.
-export function resolveHome({ isAdmin, distributed, p2pEnabled }) {
-  if (!isAdmin) return '/app/chat'
-  if (distributed) return '/app/nodes'
-  if (p2pEnabled) return '/app/p2p'
-  return ''
-}

core/http/react-ui/src/utils/sidebarPolicy.js

@@ -1,20 +0,0 @@
-import { operateConsole, isConsoleItemVisible } from '../components/console/consoleConfig'
-
-// The Operate > Cluster group, surfaced as a pinned top-of-sidebar quick-access
-// group when the admin is running a cluster (NATS-distributed) or a P2P swarm.
-// Items are filtered through the SAME gate as everywhere else, so e.g. in a
-// p2p-only deployment Nodes/Scheduling (feature: 'distributed') drop out and
-// only Swarm remains. Returns [] when the pin does not apply.
-export function clusterPinItems(auth, deployment) {
-  if (!auth.isAdmin) return []
-  if (!deployment.distributed && !deployment.p2pEnabled) return []
-  const group = operateConsole.groups.find(g => g.titleKey === 'operate.cluster')
-  if (!group) return []
-  return group.items.filter(item => isConsoleItemVisible(item, auth))
-}
-
-// In the cluster cells the Create group defaults collapsed so the pinned
-// cluster group leads. Users can still expand it; their stored choice wins.
-export function shouldCollapseCreate(auth, deployment) {
-  return !!auth.isAdmin && (!!deployment.distributed || !!deployment.p2pEnabled)
-}

core/services/distributed/gallery.go

@@ -79,21 +79,29 @@ func (s *GalleryStore) Create(op *GalleryOperationRecord) error {
 	}).Create(op).Error
 }
 
-// UpdateProgress updates progress for an operation.
-func (s *GalleryStore) UpdateProgress(id string, progress float64, message, downloadedSize string) error {
+// UpdateProgress updates progress for an operation. The cancellable flag is
+// persisted on every tick so a replica that restarts mid-install rehydrates the
+// op as still cancellable — otherwise the column keeps its Create-time zero
+// value (false), the UI hides the cancel button, and the orphaned op can only
+// be dismissed by waiting for the 30-minute stale reaper.
+func (s *GalleryStore) UpdateProgress(id string, progress float64, message, downloadedSize string, cancellable bool) error {
 	return s.db.Model(&GalleryOperationRecord{}).Where("id = ?", id).Updates(map[string]any{
 		"progress":             progress,
 		"message":              message,
 		"downloaded_file_size": downloadedSize,
+		"cancellable":          cancellable,
 		"updated_at":           time.Now(),
 	}).Error
 }
 
-// UpdateStatus updates the status of an operation.
+// UpdateStatus updates the status of an operation. A terminal status is never
+// cancellable, so the flag is cleared here to keep the persisted row consistent
+// with what the UI should offer.
 func (s *GalleryStore) UpdateStatus(id, status, errMsg string) error {
 	updates := map[string]any{
-		"status":     status,
-		"updated_at": time.Now(),
+		"status":      status,
+		"cancellable": false,
+		"updated_at":  time.Now(),
 	}
 	if errMsg != "" {
 		updates["error"] = errMsg

core/services/galleryop/cancellable_persist_test.go

@@ -0,0 +1,56 @@
+package galleryop_test
+
+import (
+	. "github.com/onsi/ginkgo/v2"
+	. "github.com/onsi/gomega"
+
+	"github.com/mudler/LocalAI/core/config"
+	"github.com/mudler/LocalAI/core/services/distributed"
+	"github.com/mudler/LocalAI/core/services/galleryop"
+	"github.com/mudler/LocalAI/core/services/testutil"
+)
+
+// Reproduces "an in-flight install can't be cancelled after a restart". The
+// live install path marks OpStatus.Cancellable=true on every progress tick, but
+// UpdateStatus persisted progress/status to the gallery store WITHOUT the
+// cancellable flag, and Create defaulted it to false. So after a replica
+// restart Hydrate rebuilt the op with Cancellable=false, /api/operations
+// reported cancellable:false, and the UI hid the cancel button — the orphaned
+// op lingered until the 30-minute stale reaper expired it. The cancellable
+// state must be persisted so a rehydrated in-flight op stays cancellable.
+var _ = Describe("GalleryService cancellable persistence across restart", func() {
+	It("rehydrates an in-flight op as still cancellable", func() {
+		db := testutil.SetupTestDB()
+		store, err := distributed.NewGalleryStore(db)
+		Expect(err).ToNot(HaveOccurred())
+
+		svc := galleryop.NewGalleryService(&config.ApplicationConfig{}, nil)
+		svc.SetGalleryStore(store)
+
+		// Seed the in-flight op row as the worker goroutine does on admission.
+		Expect(store.Create(&distributed.GalleryOperationRecord{
+			ID:                 "op-inflight",
+			GalleryElementName: "llama-cpp-development",
+			OpType:             "backend_install",
+			Status:             "pending",
+		})).To(Succeed())
+
+		// Simulate a progress tick: the live path always marks installs
+		// cancellable while they are downloading/processing.
+		svc.UpdateStatus("op-inflight", &galleryop.OpStatus{
+			Message:     "downloading",
+			Progress:    25,
+			Cancellable: true,
+		})
+
+		// A fresh replica boots and hydrates from the store.
+		fresh := galleryop.NewGalleryService(&config.ApplicationConfig{}, nil)
+		fresh.SetGalleryStore(store)
+		Expect(fresh.Hydrate()).To(Succeed())
+
+		st := fresh.GetStatus("op-inflight")
+		Expect(st).ToNot(BeNil(), "the in-flight op must hydrate after a restart")
+		Expect(st.Cancellable).To(BeTrue(),
+			"a still-active install must rehydrate as cancellable so the admin can dismiss it")
+	})
+})

core/services/galleryop/service.go

@@ -167,7 +167,7 @@ func (g *GalleryService) UpdateStatus(s string, op *OpStatus) {
 				xlog.Warn("Failed to persist gallery operation status", "op_id", s, "error", err)
 			}
 		} else {
-			if err := store.UpdateProgress(s, op.Progress, op.Message, op.DownloadedFileSize); err != nil {
+			if err := store.UpdateProgress(s, op.Progress, op.Message, op.DownloadedFileSize, op.Cancellable); err != nil {
 				xlog.Warn("Failed to persist gallery operation progress", "op_id", s, "error", err)
 			}
 		}
@@ -467,6 +467,7 @@ func (g *GalleryService) Start(c context.Context, cl *config.ModelConfigLoader,
 						GalleryElementName: op.GalleryElementName,
 						OpType:             "backend_install",
 						Status:             "pending",
+						Cancellable:        true,
 					})
 				}
 				err := g.backendHandler(&op, systemState)
@@ -499,6 +500,8 @@ func (g *GalleryService) Start(c context.Context, cl *config.ModelConfigLoader,
 						GalleryElementName: op.GalleryElementName,
 						OpType:             opType,
 						Status:             "pending",
+						// A delete is not cancellable; an install is.
+						Cancellable: !op.Delete,
 					})
 				}
 				err := g.modelHandler(&op, cl, systemState)

gallery/index.yaml

@@ -1,4 +1,175 @@
 ---
+- name: "qwopus3.6-27b-coder-compat-mtp"
+  url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
+  urls:
+    - https://huggingface.co/Jackrong/Qwopus3.6-27B-Coder-Compat-MTP-GGUF
+  description: "\U0001FA90 Qwopus-3.6-27B-Coder\nCoder SFT Release\n\nAgentic Coding & Tool-Use Reasoning Model Fine-Tuned on Qwopus3.6-27B-v2\n\n\U0001F9EC Trace Inversion & Negentropy\n\U0001F9E0 27B Dense Model\n⚡ Agentic Coding\n\U0001F6E0️ Tool Calling & Agent\n\U0001F3C6 SWE-bench Verified: 67.0% (off-thinking)\n\n\U0001F4A1 What is Qwopus-3.6-27B-Coder?\n\U0001FA90 Qwopus-3.6-27B-Coder is a reasoning-enhanced agentic coding model built on top of Qwopus3.6-27B-v2. It inherits the powerful reasoning foundation of the v2 base — which achieved 87.43% MMLU-Pro and 75.25% SWE-bench Verified — and further specializes it for agentic code generation, structured tool calling, debugging, and instruction-following in developer workflows. The model is designed to excel at repository-level coding tasks, multi-turn tool orchestration, and complex logical reasoning under realistic agent environments.\n\n\U0001F9E9 Agentic Coding\nOptimized for repository-level coding, debugging, patch generation, and structured multi-step development workflows.\n\n\U0001F6E0️ Tool Calling\nLearns from real agent trajectories with tool definitions, tool calls, and environment feedback for robust multi-turn execution.\n\n...\n"
+  license: "apache-2.0"
+  tags:
+    - llm
+    - gguf
+    - vision
+    - multimodal
+    - reasoning
+  icon: https://cdn-uploads.huggingface.co/production/uploads/66309bd090589b7c65950665/sGQKmrMc6L6guMoaB5_Y2.png
+  overrides:
+    backend: llama-cpp
+    function:
+      automatic_tool_parsing_fallback: true
+      grammar:
+        disable: true
+    known_usecases:
+      - chat
+    mmproj: llama-cpp/mmproj/Qwopus3.6-27B-Coder-Compat-MTP-GGUF/mmproj-F32.gguf
+    options:
+      - use_jinja:true
+      - spec_type:draft-mtp
+      - spec_n_max:6
+      - spec_p_min:0.75
+    parameters:
+      model: llama-cpp/models/Qwopus3.6-27B-Coder-Compat-MTP-GGUF/Qwopus3.6-27B-Coder-Compat-MTP-Q4_K_M.gguf
+    template:
+      use_tokenizer_template: true
+  files:
+    - filename: llama-cpp/models/Qwopus3.6-27B-Coder-Compat-MTP-GGUF/Qwopus3.6-27B-Coder-Compat-MTP-Q4_K_M.gguf
+      sha256: f893632170124da60e159b7bcc9d91e1cda3014b2c6b8ad9c6cde38a1fcd2f6f
+      uri: https://huggingface.co/Jackrong/Qwopus3.6-27B-Coder-Compat-MTP-GGUF/resolve/main/Qwopus3.6-27B-Coder-Compat-MTP-Q4_K_M.gguf
+    - filename: llama-cpp/mmproj/Qwopus3.6-27B-Coder-Compat-MTP-GGUF/mmproj-F32.gguf
+      sha256: 32f7ea0600c07272547da401d460f8abbd980f3a57b69d6df87be0e2505e0b9c
+      uri: https://huggingface.co/Jackrong/Qwopus3.6-27B-Coder-Compat-MTP-GGUF/resolve/main/mmproj-F32.gguf
+- name: "kimi-k2.7-code"
+  url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
+  urls:
+    - https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF
+  description: |
+    ## 1. Model Introduction
+
+    Kimi K2.7 Code is a coding-focused agentic model built upon Kimi K2.6. With substantial improvements on real-world long-horizon coding tasks, it strengthens end-to-end task completion across complex software engineering workflows while improving token efficiency, reducing thinking-token usage by approximately 30% compared with Kimi K2.6.
+
+    ## 2. Model Summary
+
+    ## 3. Evaluation Results
+
+    Benchmark
+    Kimi K2.6
+    Kimi K2.7 Code
+    GPT-5.5
+    Claude Opus 4.8
+
+    Coding
+
+    Kimi Code Bench v2
+    50.9
+    62.0
+    69.0
+    67.4
+
+    Program Bench
+    48.3
+    53.6
+    69.1
+    63.8
+
+    MLS Bench Lite
+    26.7
+    35.1
+    35.5
+    42.8
+
+    Agentic
+
+    Kimi Claw 24/7 Bench
+    42.9
+    46.9
+    52.8
+    50.4
+
+    MCP Atlas
+    69.4
+    76.0
+    79.4
+    81.3
+
+    MCP Mark Verified
+    72.8
+    81.1
+    92.9
+    76.4
+
+    Footnotes
+
+    ...
+  license: "other"
+  tags:
+    - llm
+    - gguf
+  icon: https://huggingface.co/moonshotai/Kimi-K2.7-Code/resolve/main/figures/kimi-logo.png
+  overrides:
+    backend: llama-cpp
+    function:
+      automatic_tool_parsing_fallback: true
+      grammar:
+        disable: true
+    known_usecases:
+      - chat
+    mmproj: llama-cpp/mmproj/Kimi-K2.7-Code-GGUF/mmproj-F32.gguf
+    options:
+      - use_jinja:true
+    parameters:
+      min_p: 0.01
+      model: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00001-of-00014.gguf
+      repeat_penalty: 1
+      temperature: 0.6
+      top_k: -1
+      top_p: 0.95
+    template:
+      use_tokenizer_template: true
+  files:
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00001-of-00014.gguf
+      sha256: 65f0aca336f876902323a90e2aff32cac76d071b2cdd818c6a8d78be8fc2c680
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00001-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00002-of-00014.gguf
+      sha256: 40f4416c130827a11502778891f4ef95b2144db90f51d63aa3548d0952a39683
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00002-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00003-of-00014.gguf
+      sha256: ba2ba0b5168784ace7c752ecadfc3631279b2bb023824cb0fe9e2dab3dd28f22
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00003-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00004-of-00014.gguf
+      sha256: 10298a6c98b13ef49be286fefbea8663e16473fb69bbeabe153bc80c60ae116e
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00004-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00005-of-00014.gguf
+      sha256: 8e9e4c8e35d34fc4fef6bfb65a715ad7defbd196970d833c1df6924d701c88b3
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00005-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00006-of-00014.gguf
+      sha256: ccff6e7f299742f82cf6f51a871e3eb3167511efaee967477cc8387f54d16442
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00006-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00007-of-00014.gguf
+      sha256: 1a3b639633a2d22f71156a9f643ded2329cdd969cc21177b644b5741bac1af8e
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00007-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00008-of-00014.gguf
+      sha256: bde28f682a1eab973538b2102007d952f37a13c1f7d55e2ed99177445ddc4282
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00008-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00009-of-00014.gguf
+      sha256: b6a23a95b61e100f7593fa75e2363966323fa767b7e4fdf45d963b59e8fdc69f
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00009-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00010-of-00014.gguf
+      sha256: fb10231c2e6d76921d40f22690f4aa08a8090c708edeaf7e581abafc24d3b25c
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00010-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00011-of-00014.gguf
+      sha256: d2290be7ed1a22ac1f9f8a4813389689e075ce2ab8abc3aaaa1157a3cb1462d8
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00011-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00012-of-00014.gguf
+      sha256: ce0d028314aa3fc783082dbca097e1055d69686a17ab8306574e2949568f26a5
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00012-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00013-of-00014.gguf
+      sha256: 217864ce63a1d130ab39dcb0996b6097e1aa78eb896e38efaefdbbac3a00b7ec
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00013-of-00014.gguf
+    - filename: llama-cpp/models/Kimi-K2.7-Code-GGUF/Kimi-K2.7-Code-UD-Q8_K_XL-00014-of-00014.gguf
+      sha256: eb7582ad7066c5eaa01bde95acb00b4ad9cd7b07cd50a6cf5c9ee427258bc9dd
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/UD-Q8_K_XL/Kimi-K2.7-Code-UD-Q8_K_XL-00014-of-00014.gguf
+    - filename: llama-cpp/mmproj/Kimi-K2.7-Code-GGUF/mmproj-F32.gguf
+      sha256: b2cc50c8c13fe70fc4968a83332f31e9007ea09ebb9ae91d46a4e4cd2a3053cd
+      uri: https://huggingface.co/unsloth/Kimi-K2.7-Code-GGUF/resolve/main/mmproj-F32.gguf
 - name: "qwythos-9b-claude-mythos-5-1m"
   url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
   urls:
@@ -49,33 +220,7 @@
   url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
   urls:
     - https://huggingface.co/unsloth/GLM-5.2-GGUF
-  description: |
-    # GLM-5.2
-
-    👋 Join our WeChat or Discord community.
-
-    📖 Check out the GLM-5.2 blog and GLM-5 Technical report.
-
-    📍 Use GLM-5.2 API services on Z.ai API Platform.
-
-    🔜 Try GLM-5.2 here.
-
-    [Paper]
-    [GitHub]
-
-    ## Introduction
-
-    We're introducing GLM-5.2, our latest flagship model for long-horizon tasks. It marks a substantial leap in long-horizon task capability over its predecessor GLM-5.1 and, for the first time, delivers that capability on a **solid 1M-token context**. GLM-5.2's new capabilities include:
-      - **Solid 1M Context:** A solid 1M-token context that stably sustains long-horizon work
-      - **Advanced Coding with Flexible Effort**: Stronger coding capabilities with multiple thinking effort levels to balance performance and latency
-      - **Improved Architecture**: We propose IndexShare, which reuses the same indexer across every four sparse attention layers, reducing per-token FLOPs by 2.9× at a 1M context length. We also improve GLM-5.2’s MTP layer for speculative decoding, increasing the acceptance length by up to 20%
-      - **Pure Open**: An MIT open-source license — no regional limits, technical access without borders
-
-    ## Benchmark
-
-    ## Serve GLM-5.2 Locally
-
-    ...
+  description: "# GLM-5.2\n\n\U0001F44B Join our WeChat or Discord community.\n\n\U0001F4D6 Check out the GLM-5.2 blog and GLM-5 Technical report.\n\n\U0001F4CD Use GLM-5.2 API services on Z.ai API Platform.\n\n\U0001F51C Try GLM-5.2 here.\n\n[Paper]\n[GitHub]\n\n## Introduction\n\nWe're introducing GLM-5.2, our latest flagship model for long-horizon tasks. It marks a substantial leap in long-horizon task capability over its predecessor GLM-5.1 and, for the first time, delivers that capability on a **solid 1M-token context**. GLM-5.2's new capabilities include:\n  - **Solid 1M Context:** A solid 1M-token context that stably sustains long-horizon work\n  - **Advanced Coding with Flexible Effort**: Stronger coding capabilities with multiple thinking effort levels to balance performance and latency\n  - **Improved Architecture**: We propose IndexShare, which reuses the same indexer across every four sparse attention layers, reducing per-token FLOPs by 2.9× at a 1M context length. We also improve GLM-5.2’s MTP layer for speculative decoding, increasing the acceptance length by up to 20%\n  - **Pure Open**: An MIT open-source license — no regional limits, technical access without borders\n\n## Benchmark\n\n## Serve GLM-5.2 Locally\n\n...\n"
   license: "mit"
   tags:
     - llm
@@ -198,26 +343,7 @@
   url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
   urls:
     - https://huggingface.co/michaelw9999/Qwopus3.6-27B-v2-MTP-NVFP4-GGUF
-  description: |
-    🪐 Qwopus3.6-27B-v2-MTP
-    MTP Release
-
-    Multi-Token Prediction reasoning model fine-tuned from Qwen3.6-27B
-
-    🧬 Trace Inversion & Negentropy
-    🧠 27B Parameters
-    ⚡ Speculative Decoding
-    🛠️ Coding / DevOps / Math
-
-    💡 What is Qwopus3.6-27B-v2-MTP?
-    🪐 Qwopus3.6-27B-v2-MTP is a speed-oriented reasoning release built on top of Qwen3.6-27B. It keeps the Qwopus line's focus on reconstructed reasoning traces, coding discipline, DevOps procedures, and mathematical derivations, while adding Multi-Token Prediction for faster generation. The goal is simple: preserve the depth and structure of a 27B reasoning model while making real interactive use noticeably faster.
-
-    ⚡ MTP DecodingAuxiliary future-token prediction improves throughput on long reasoning, code, math, and strict-format prompts.
-    🧩 Structured ReasoningInherits the Qwopus training recipe built around reconstructed step-by-step reasoning trajectories.
-    🧪 GB10 TestedValidated on a 30-question local benchmark across Logic, Coding, DevOps, Math, and Edge tasks.
-    🚀 Practical SpeedDesigned for workflows where strong answers matter, but waiting several extra minutes per task does not.
-
-    ...
+  description: "\U0001FA90 Qwopus3.6-27B-v2-MTP\nMTP Release\n\nMulti-Token Prediction reasoning model fine-tuned from Qwen3.6-27B\n\n\U0001F9EC Trace Inversion & Negentropy\n\U0001F9E0 27B Parameters\n⚡ Speculative Decoding\n\U0001F6E0️ Coding / DevOps / Math\n\n\U0001F4A1 What is Qwopus3.6-27B-v2-MTP?\n\U0001FA90 Qwopus3.6-27B-v2-MTP is a speed-oriented reasoning release built on top of Qwen3.6-27B. It keeps the Qwopus line's focus on reconstructed reasoning traces, coding discipline, DevOps procedures, and mathematical derivations, while adding Multi-Token Prediction for faster generation. The goal is simple: preserve the depth and structure of a 27B reasoning model while making real interactive use noticeably faster.\n\n⚡ MTP DecodingAuxiliary future-token prediction improves throughput on long reasoning, code, math, and strict-format prompts.\n\U0001F9E9 Structured ReasoningInherits the Qwopus training recipe built around reconstructed step-by-step reasoning trajectories.\n\U0001F9EA GB10 TestedValidated on a 30-question local benchmark across Logic, Coding, DevOps, Math, and Edge tasks.\n\U0001F680 Practical SpeedDesigned for workflows where strong answers matter, but waiting several extra minutes per task does not.\n\n...\n"
   tags:
     - llm
     - gguf
@@ -243,28 +369,7 @@
   url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
   urls:
     - https://huggingface.co/michaelw9999/Qwopus3.6-27B-Coder-MTP-NVFP4-GGUF
-  description: |
-    🪐 Qwopus-3.6-27B-Coder
-    Coder SFT Release
-
-    Agentic Coding & Tool-Use Reasoning Model Fine-Tuned on Qwopus3.6-27B-v2
-
-    🧬 Trace Inversion & Negentropy
-    🧠 27B Dense Model
-    ⚡ Agentic Coding
-    🛠️ Tool Calling & Agent
-    🏆 SWE-bench Verified: 67.0% (off-thinking)
-
-    💡 What is Qwopus-3.6-27B-Coder?
-    🪐 Qwopus-3.6-27B-Coder is a reasoning-enhanced agentic coding model built on top of Qwopus3.6-27B-v2. It inherits the powerful reasoning foundation of the v2 base — which achieved 87.43% MMLU-Pro (300ex) and 75.25% SWE-bench Verified — and further specializes it for agentic code generation, structured tool calling, debugging, and instruction-following in developer workflows. The model is designed to excel at repository-level coding tasks, multi-turn tool orchestration, and complex logical reasoning under realistic agent environments.
-
-    🧩 Agentic Coding
-    Optimized for repository-level coding, debugging, patch generation, and structured multi-step development workflows.
-
-    🛠️ Tool Calling
-    Learns from real agent trajectories with tool definitions, tool calls, and environment feedback for robust multi-turn execution.
-
-    ...
+  description: "\U0001FA90 Qwopus-3.6-27B-Coder\nCoder SFT Release\n\nAgentic Coding & Tool-Use Reasoning Model Fine-Tuned on Qwopus3.6-27B-v2\n\n\U0001F9EC Trace Inversion & Negentropy\n\U0001F9E0 27B Dense Model\n⚡ Agentic Coding\n\U0001F6E0️ Tool Calling & Agent\n\U0001F3C6 SWE-bench Verified: 67.0% (off-thinking)\n\n\U0001F4A1 What is Qwopus-3.6-27B-Coder?\n\U0001FA90 Qwopus-3.6-27B-Coder is a reasoning-enhanced agentic coding model built on top of Qwopus3.6-27B-v2. It inherits the powerful reasoning foundation of the v2 base — which achieved 87.43% MMLU-Pro (300ex) and 75.25% SWE-bench Verified — and further specializes it for agentic code generation, structured tool calling, debugging, and instruction-following in developer workflows. The model is designed to excel at repository-level coding tasks, multi-turn tool orchestration, and complex logical reasoning under realistic agent environments.\n\n\U0001F9E9 Agentic Coding\nOptimized for repository-level coding, debugging, patch generation, and structured multi-step development workflows.\n\n\U0001F6E0️ Tool Calling\nLearns from real agent trajectories with tool definitions, tool calls, and environment feedback for robust multi-turn execution.\n\n...\n"
   tags:
     - llm
     - gguf
@@ -1484,8 +1589,8 @@
       use_tokenizer_template: true
   files:
     - filename: llama-cpp/models/Qwopus3.6-27B-v2-MTP-GGUF/Qwopus3.6-27B-v2-MTP-Q4_K_M.gguf
-      sha256: 818d68223be4d8518dac0b3b5604dde633cbbcbae1f491d842a3e26711c6606d
       uri: https://huggingface.co/Jackrong/Qwopus3.6-27B-v2-MTP-GGUF/resolve/main/Qwopus3.6-27B-v2-MTP-Q4_K_M.gguf
+      sha256: 31cf5fc2406a0c7aaebcc26d440bf0df94e215d0589d5205bf319649c052b50a
 - name: "qwen3.6-40b-claude-4.6-opus-deckard-heretic-uncensored-thinking-neo-code-di-imatrix-max"
   url: "github:mudler/LocalAI/gallery/virtual.yaml@master"
   urls:

tests/e2e/distributed/gallery_distributed_test.go

@@ -53,12 +53,13 @@ var _ = Describe("Gallery Distributed", Label("Distributed"), func() {
 			Expect(retrieved.Status).To(Equal("downloading"))
 			Expect(retrieved.FrontendID).To(Equal("f1"))
 
-			// Update progress
-			Expect(galleryStore.UpdateProgress(op.ID, 0.75, "75% complete", "6GB")).To(Succeed())
+			// Update progress (cancellable: a downloading install can be cancelled)
+			Expect(galleryStore.UpdateProgress(op.ID, 0.75, "75% complete", "6GB", true)).To(Succeed())
 
 			updated, _ := galleryStore.Get(op.ID)
 			Expect(updated.Progress).To(BeNumerically("~", 0.75, 0.01))
 			Expect(updated.Message).To(Equal("75% complete"))
+			Expect(updated.Cancellable).To(BeTrue())
 
 			// Complete
 			Expect(galleryStore.UpdateStatus(op.ID, "completed", "")).To(Succeed())

tests/e2e/distributed/phase4_test.go

@@ -104,11 +104,12 @@ var _ = Describe("Phase 4: MCP, Skills, Gallery, Fine-Tuning", Label("Distribute
 			}
 			stores.Gallery.Create(op)
 
-			Expect(stores.Gallery.UpdateProgress(op.ID, 0.5, "50% complete", "2GB")).To(Succeed())
+			Expect(stores.Gallery.UpdateProgress(op.ID, 0.5, "50% complete", "2GB", true)).To(Succeed())
 
 			updated, _ := stores.Gallery.Get(op.ID)
 			Expect(updated.Progress).To(BeNumerically("~", 0.5, 0.01))
 			Expect(updated.Message).To(Equal("50% complete"))
+			Expect(updated.Cancellable).To(BeTrue())
 		})
 
 		It("should deduplicate concurrent downloads", func() {