diff --git a/admin/app/controllers/settings_controller.ts b/admin/app/controllers/settings_controller.ts
index c21ddd6..24cb4ce 100644
--- a/admin/app/controllers/settings_controller.ts
+++ b/admin/app/controllers/settings_controller.ts
@@ -64,6 +64,7 @@ export default class SettingsController {
     const chatSuggestionsEnabled = await KVStore.getValue('chat.suggestionsEnabled')
     const aiAssistantCustomName = await KVStore.getValue('ai.assistantCustomName')
     const remoteOllamaUrl = await KVStore.getValue('ai.remoteOllamaUrl')
+    const ollamaFlashAttention = await KVStore.getValue('ai.ollamaFlashAttention')
     return inertia.render('settings/models', {
       models: {
         availableModels: availableModels?.models || [],
@@ -72,6 +73,7 @@ export default class SettingsController {
           chatSuggestionsEnabled: chatSuggestionsEnabled ?? false,
           aiAssistantCustomName: aiAssistantCustomName ?? '',
           remoteOllamaUrl: remoteOllamaUrl ?? '',
+          ollamaFlashAttention: ollamaFlashAttention ?? true,
         },
       },
     })
diff --git a/admin/app/services/docker_service.ts b/admin/app/services/docker_service.ts
index d3edb5d..3c44332 100644
--- a/admin/app/services/docker_service.ts
+++ b/admin/app/services/docker_service.ts
@@ -505,6 +505,15 @@ export class DockerService {
         }
       }
 
+      const ollamaEnv: string[] = []
+      if (service.service_name === SERVICE_NAMES.OLLAMA) {
+        ollamaEnv.push('OLLAMA_NO_CLOUD=1')
+        const flashAttentionEnabled = await KVStore.getValue('ai.ollamaFlashAttention')
+        if (flashAttentionEnabled !== false) {
+          ollamaEnv.push('OLLAMA_FLASH_ATTENTION=1')
+        }
+      }
+
       this._broadcast(
         service.service_name,
         'creating',
@@ -522,7 +531,7 @@ export class DockerService {
         HostConfig: gpuHostConfig,
         ...(containerConfig?.WorkingDir && { WorkingDir: containerConfig.WorkingDir }),
         ...(containerConfig?.ExposedPorts && { ExposedPorts: containerConfig.ExposedPorts }),
-        ...(containerConfig?.Env && { Env: containerConfig.Env }),
+        Env: [...(containerConfig?.Env ?? []), ...ollamaEnv],
         ...(service.container_command ? { Cmd: service.container_command.split(' ') } : {}),
         // Ensure container is attached to the Nomad docker network in production
         ...(process.env.NODE_ENV === 'production' && {
diff --git a/admin/constants/kv_store.ts b/admin/constants/kv_store.ts
index 1085723..c49416c 100644
--- a/admin/constants/kv_store.ts
+++ b/admin/constants/kv_store.ts
@@ -1,3 +1,3 @@
 import { KVStoreKey } from "../types/kv_store.js";
 
-export const SETTINGS_KEYS: KVStoreKey[] = ['chat.suggestionsEnabled', 'chat.lastModel', 'ui.hasVisitedEasySetup', 'ui.theme', 'system.earlyAccess', 'ai.assistantCustomName', 'ai.remoteOllamaUrl'];
\ No newline at end of file
+export const SETTINGS_KEYS: KVStoreKey[] = ['chat.suggestionsEnabled', 'chat.lastModel', 'ui.hasVisitedEasySetup', 'ui.theme', 'system.earlyAccess', 'ai.assistantCustomName', 'ai.remoteOllamaUrl', 'ai.ollamaFlashAttention'];
\ No newline at end of file
diff --git a/admin/inertia/pages/settings/models.tsx b/admin/inertia/pages/settings/models.tsx
index 08d9616..d4124fe 100644
--- a/admin/inertia/pages/settings/models.tsx
+++ b/admin/inertia/pages/settings/models.tsx
@@ -26,7 +26,7 @@ export default function ModelsPage(props: {
   models: {
     availableModels: NomadOllamaModel[]
     installedModels: NomadInstalledModel[]
-    settings: { chatSuggestionsEnabled: boolean; aiAssistantCustomName: string; remoteOllamaUrl: string }
+    settings: { chatSuggestionsEnabled: boolean; aiAssistantCustomName: string; remoteOllamaUrl: string; ollamaFlashAttention: boolean }
   }
 }) {
   const { aiAssistantName } = usePage<{ aiAssistantName: string }>().props
@@ -95,6 +95,9 @@ export default function ModelsPage(props: {
   const [chatSuggestionsEnabled, setChatSuggestionsEnabled] = useState(
     props.models.settings.chatSuggestionsEnabled
   )
+  const [ollamaFlashAttention, setOllamaFlashAttention] = useState(
+    props.models.settings.ollamaFlashAttention
+  )
   const [aiAssistantCustomName, setAiAssistantCustomName] = useState(
     props.models.settings.aiAssistantCustomName
   )
@@ -308,6 +311,15 @@ export default function ModelsPage(props: {
                 label="Chat Suggestions"
                 description="Display AI-generated conversation starters in the chat interface"
               />
+              <Switch
+                checked={ollamaFlashAttention}
+                onChange={(newVal) => {
+                  setOllamaFlashAttention(newVal)
+                  updateSettingMutation.mutate({ key: 'ai.ollamaFlashAttention', value: newVal })
+                }}
+                label="Flash Attention"
+                description="Enables OLLAMA_FLASH_ATTENTION=1 for improved memory efficiency. Disable if you experience instability. Takes effect after reinstalling the AI Assistant."
+              />
               <Input
                 name="aiAssistantCustomName"
                 label="Assistant Name"
diff --git a/admin/types/kv_store.ts b/admin/types/kv_store.ts
index ec568c3..8fb2686 100644
--- a/admin/types/kv_store.ts
+++ b/admin/types/kv_store.ts
@@ -11,6 +11,7 @@ export const KV_STORE_SCHEMA = {
   'ai.assistantCustomName':     'string',
   'gpu.type':                   'string',
   'ai.remoteOllamaUrl':         'string',
+  'ai.ollamaFlashAttention':    'boolean',
 } as const
 
 type KVTagToType<T extends string> = T extends 'boolean' ? boolean : string