Update ollama to version v0.19.0 (#5237)

ollama/docker-compose.yml

@@ -8,7 +8,7 @@ services:
       PROXY_AUTH_ADD: "false"
 
   ollama:
-    image: ollama/ollama:0.18.3@sha256:e305a7b6529c3deada679f07f98ac3222ce668bfc970062f83a5e4e7e15b8f6d
+    image: ollama/ollama:0.19.0@sha256:bf240c2847a8bc7b2c630b85dab5d1dedcba257b551d5fc9b290ce544d59272a
     environment:
       OLLAMA_ORIGINS: "*"
       OLLAMA_CONTEXT_LENGTH: 8192

ollama/umbrel-app.yml

@@ -3,7 +3,7 @@ id: ollama
 name: Ollama
 tagline: Self-host open source AI models like DeepSeek-R1, Llama, and more
 category: ai
-version: "0.18.3"
+version: "0.19.0"
 port: 11434
 description: >-
   Ollama allows you to download and run advanced AI models directly on your own hardware. Self-hosting AI models ensures full control over your data and protects your privacy.
@@ -38,13 +38,12 @@ defaultUsername: ""
 defaultPassword: ""
 dependencies: []
 releaseNotes: >-
-  This release adds Visual Studio Code integration and improvements to OpenClaw.
-
-
   Key highlights in this release:
-    - Microsoft Visual Studio Code now directly integrates with Ollama via GitHub Copilot, allowing any local or cloud model from Ollama to be used within VS Code
-    - GLM parser improvements for tool calls
-    - OpenClaw integration improvements for gateway checks
+    - `ollama launch pi` now includes a web search plugin that uses Ollama's web search
+    - Improved KV cache hit rate when using the Anthropic-compatible API
+    - Fixed tool call parsing issue with Qwen3.5 where tool calls would appear in thinking output
+    - Fixed issue where flash attention would be incorrectly enabled for `grok` models
+    - Fixed `qwen3-next:80b` not loading in Ollama
 
 
   Full release notes are available at https://github.com/ollama/ollama/releases