diff --git a/ollamagemma431a4budq4kxlv2/Chart.yaml b/ollamagemma431a4budq4kxlv2/Chart.yaml
index f54c9a5a8..05b4d3075 100644
--- a/ollamagemma431a4budq4kxlv2/Chart.yaml
+++ b/ollamagemma431a4budq4kxlv2/Chart.yaml
@@ -3,4 +3,4 @@ appVersion: 'gemma4:31b-it-ud-q4_K_XL'
 description: description
 name: ollamagemma431a4budq4kxlv2
 type: application
-version: '1.0.9'
\ No newline at end of file
+version: '1.0.10'
diff --git a/ollamagemma431a4budq4kxlv2/OlaresManifest.yaml b/ollamagemma431a4budq4kxlv2/OlaresManifest.yaml
index b5a343c2e..76dc1f153 100644
--- a/ollamagemma431a4budq4kxlv2/OlaresManifest.yaml
+++ b/ollamagemma431a4budq4kxlv2/OlaresManifest.yaml
@@ -8,7 +8,7 @@ metadata:
   description: "Gemma4 31B IT dense (Unsloth Dynamic UD-Q4_K_XL) for efficient local inference"
   appid: ollamagemma431a4budq4kxlv2
   title: Gemma4 31B IT UD-XL (Ollama)
-  version: '1.0.9'
+  version: '1.0.10'
   categories:
     - AI
 sharedEntrances:
@@ -46,8 +46,8 @@ spec:
     - Context: 65536 (safe on 24GB; increase if VRAM allows)
 
   upgradeDescription: |
-    Upgrade Ollama runtime to ollama/ollama:0.30.10.
-    Upgrade Olares API proxy to beclab/harveyff-olares-ollama:v0.1.22 (OpenAI vision image_url forwarding).
+    Context length raised to 65536 (safe on 24GB with q8_0 KV). Thinking enabled by default; clients can disable it per request. Vision support via `mmproj-F16.gguf`, with tool calling. Proxy v0.1.21, Ollama 0.30.0. KV cache q8_0.
+
   developer: Google (base) / Unsloth (GGUF)
   website: https://ai.google.dev/gemma
   submitter: Olares
@@ -99,7 +99,7 @@ options:
   {{- end }}
   dependencies:
     - name: olares
-      version: '>=1.12.3-0'
+      version: '>=1.12.3-0,<1.12.6'
       type: system
   {{- if and .Values.admin .Values.bfl.username (eq .Values.admin .Values.bfl.username) }}
   {{- else }}
diff --git a/ollamagemma431a4budq4kxlv2/i18n/en-US/OlaresManifest.yaml b/ollamagemma431a4budq4kxlv2/i18n/en-US/OlaresManifest.yaml
index 9a4c7100c..92119d485 100644
--- a/ollamagemma431a4budq4kxlv2/i18n/en-US/OlaresManifest.yaml
+++ b/ollamagemma431a4budq4kxlv2/i18n/en-US/OlaresManifest.yaml
@@ -19,5 +19,4 @@ spec:
     - Context: 32768
 
   upgradeDescription: |
-    Upgrade Ollama runtime to ollama/ollama:0.30.10.
-    Upgrade Olares API proxy to beclab/harveyff-olares-ollama:v0.1.22 (OpenAI vision image_url forwarding).
\ No newline at end of file
+    Context raised to 65536 (safe on 24GB). Add vision: `mmproj-F16.gguf` + embedded chat template for tools/vision. Thinking on by default (OLLAMA_THINKING=""). Proxy v0.1.21, Ollama 0.30.0. KV cache q8_0.
diff --git a/ollamagemma431a4budq4kxlv2/i18n/zh-CN/OlaresManifest.yaml b/ollamagemma431a4budq4kxlv2/i18n/zh-CN/OlaresManifest.yaml
index 7eb17bec1..fab94ab2d 100644
--- a/ollamagemma431a4budq4kxlv2/i18n/zh-CN/OlaresManifest.yaml
+++ b/ollamagemma431a4budq4kxlv2/i18n/zh-CN/OlaresManifest.yaml
@@ -19,5 +19,4 @@ spec:
     - 上下文：32768
 
   upgradeDescription: |
-    升级 Ollama 运行时至 ollama/ollama:0.30.10。
-    升级 Olares API 代理至 beclab/harveyff-olares-ollama:v0.1.22（OpenAI vision image_url 转发）。
+    上下文提升至 65536（24GB 安全）。新增视觉：`mmproj-F16.gguf` + 内嵌对话模板以支持 tools/vision。默认开启思考（OLLAMA_THINKING=""）。代理 v0.1.21，Ollama 0.30.0。KV cache q8_0。
diff --git a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2/Chart.yaml b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2/Chart.yaml
index a1a9d5e71..5d372ebfb 100644
--- a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2/Chart.yaml
+++ b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2/Chart.yaml
@@ -3,4 +3,4 @@ appVersion: '1.25.3-2'
 description: description
 name: ollamagemma431a4budq4kxlv2
 type: application
-version: '1.0.9'
\ No newline at end of file
+version: '1.0.10'
diff --git a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/Chart.yaml b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/Chart.yaml
index 7987a6851..23b5adcc3 100644
--- a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/Chart.yaml
+++ b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/Chart.yaml
@@ -1,6 +1,6 @@
 apiVersion: v2
-appVersion: '0.30.10'
+appVersion: '0.30.0'
 description: description
 name: ollamagemma431a4budq4kxlv2s
 type: application
-version: '1.0.9'
\ No newline at end of file
+version: '1.0.10'
diff --git a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/api.yaml b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/api.yaml
index 758717629..ae3d80410 100644
--- a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/api.yaml
+++ b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/api.yaml
@@ -28,7 +28,7 @@ spec:
     spec:
       containers:
         - name: api
-          image: "docker.io/beclab/harveyff-olares-ollama:v0.1.22"
+          image: "docker.io/beclab/harveyff-olares-ollama:v0.1.21"
           env:
             - name: PGID
               value: "1000"
diff --git a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/deployment.yaml b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/deployment.yaml
index efd20e820..875b68ac6 100644
--- a/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/deployment.yaml
+++ b/ollamagemma431a4budq4kxlv2/ollamagemma431a4budq4kxlv2s/templates/deployment.yaml
@@ -26,7 +26,7 @@ spec:
     spec:
       containers:
         - name: ollama
-          image: "docker.io/ollama/ollama:0.30.10"
+          image: "docker.io/ollama/ollama:0.30.0"
           imagePullPolicy: IfNotPresent
           command:
             - sh