diff --git a/src/modules/llms/server/openai/openai.router.ts b/src/modules/llms/server/openai/openai.router.ts
index dc5b22bc8..15ab36221 100644
--- a/src/modules/llms/server/openai/openai.router.ts
+++ b/src/modules/llms/server/openai/openai.router.ts
@@ -30,9 +30,9 @@ const _createImageConfigBase = z.object({
   user: z.string().optional(),
 });
 
-// GPT Image family (gpt-image-1.5, gpt-image-1, gpt-image-1-mini share all parameters)
+// GPT Image family (all members share the same parameter surface)
 const createImageConfigGI = _createImageConfigBase.extend({
-  model: z.enum(['gpt-image-1.5', 'gpt-image-1', 'gpt-image-1-mini']),
+  model: OpenAIWire_API_Images_Generations.GptImageModels_schema,
   prompt: z.string().max(32000),
   size: z.enum([/*'auto',*/ '1024x1024', '1536x1024', '1024x1536']),
   quality: z.enum(['high', 'medium', 'low']).optional(),
@@ -153,8 +153,8 @@ export const llmOpenAIRouter = createTRPCRouter({
 
       const { access, generationConfig: config, editConfig } = input;
 
-      // Determine if this is an edit request
-      const isGptImageFamily = config.model === 'gpt-image-1' || config.model === 'gpt-image-1-mini';
+      // Determine if this is an edit request (any member of the GPT Image family supports edits)
+      const isGptImageFamily = config.model === 'gpt-image-2' || config.model === 'gpt-image-1.5' || config.model === 'gpt-image-1' || config.model === 'gpt-image-1-mini';
       const isEdit = !!editConfig?.inputImages?.length && isGptImageFamily;
 
       // validate input
diff --git a/src/modules/t2i/dalle/DallESettings.tsx b/src/modules/t2i/dalle/DallESettings.tsx
index db444417e..0408b3f6e 100644
--- a/src/modules/t2i/dalle/DallESettings.tsx
+++ b/src/modules/t2i/dalle/DallESettings.tsx
@@ -17,6 +17,7 @@ import { openAIImageModelsPricing } from './openaiGenerateImages';
 const CONF = {
 
   MODEL_OPTS: [
+    { value: 'gpt-image-2', label: 'GPT Image 2' },
     { value: 'gpt-image-1.5', label: 'GPT Image 1.5' },
     { value: 'gpt-image-1', label: 'GPT Image 1' },
     { value: 'gpt-image-1-mini', label: 'GPT Image Mini' },
diff --git a/src/modules/t2i/dalle/openaiGenerateImages.ts b/src/modules/t2i/dalle/openaiGenerateImages.ts
index 0a8565e7a..55aad446c 100644
--- a/src/modules/t2i/dalle/openaiGenerateImages.ts
+++ b/src/modules/t2i/dalle/openaiGenerateImages.ts
@@ -85,11 +85,12 @@ export async function openAIGenerateImagesOrThrow(
       access: findServiceAccessOrThrow<{}, OpenAIAccessSchema>(modelServiceIdForAccess).transportAccess,
       // [LocalAI, 2025-11-18] LocalAI uses the default model 'stablediffusion' and we don't have any dynamic model selection yet
       generationConfig: modelVendor === 'localai' ? {
-        model: dalleModelId === 'gpt-image-1' ? 'stablediffusion'
-          : dalleModelId === 'gpt-image-1-mini' ? 'dreamshaper'
-            : dalleModelId === 'dall-e-3' ? 'sd-3.5-large-ggml'
-              : dalleModelId === 'dall-e-2' ? 'sd-3.5-medium-ggml'
-                : 'dreamshaper',
+        model: dalleModelId === 'gpt-image-2' ? 'stablediffusion'
+          : dalleModelId === 'gpt-image-1' ? 'stablediffusion'
+            : dalleModelId === 'gpt-image-1-mini' ? 'dreamshaper'
+              : dalleModelId === 'dall-e-3' ? 'sd-3.5-large-ggml'
+                : dalleModelId === 'dall-e-2' ? 'sd-3.5-medium-ggml'
+                  : 'dreamshaper',
         prompt,
         count: imageCount,
         // [LocalAI] size mapping - FIXME! - TEMP CODE
@@ -98,7 +99,7 @@ export async function openAIGenerateImagesOrThrow(
             : '1024x1024',
         response_format: 'b64_json',
       } : getImageModelFamily(dalleModelId) === 'gpt-image' ? {
-        model: dalleModelId as 'gpt-image-1.5' | 'gpt-image-1' | 'gpt-image-1-mini',
+        model: dalleModelId as 'gpt-image-2' | 'gpt-image-1.5' | 'gpt-image-1' | 'gpt-image-1-mini',
         prompt: prompt.slice(0, 32000 - 1), // GPT Image family accepts much longer prompts
         count: imageCount,
         size: dalleSizeGI,
@@ -199,6 +200,7 @@ export async function openAIGenerateImagesOrThrow(
 export function openAIImageModelsCurrentGeneratorName() {
   const dalleModelSelection = useDalleStore.getState().dalleModelId;
   const dalleModelId = resolveDalleModelId(dalleModelSelection);
+  if (dalleModelId === 'gpt-image-2') return 'GPT Image 2';
   if (dalleModelId === 'gpt-image-1.5') return 'GPT Image 1.5';
   if (dalleModelId === 'gpt-image-1') return 'GPT Image 1';
   if (dalleModelId === 'gpt-image-1-mini') return 'GPT Image Mini';
@@ -217,10 +219,13 @@ export function openAIImageModelsCurrentGeneratorName() {
  * - Deduct credits after successful generation
  */
 const IMAGE_MODEL_PRICING = {
-  // Token-based pricing (GPT Image family) - Note: chatgpt-image-latest has same pricing as gpt-image-1.5
-  'gpt-image-1.5': { inputText: 5.00, inputImage: 8.0, outputImage: 32.0 },
-  'gpt-image-1': { inputText: 5.00, inputImage: 10.0, outputImage: 40.0 },
-  'gpt-image-1-mini': { inputText: 2.00, inputImage: 2.50, outputImage: 8.00 },
+  // Token-based pricing (GPT Image family). Per $1M tokens. Note: chatgpt-image-latest mirrors gpt-image-1.5.
+  // Cached-input discounts exist (gpt-image-2/1.5: $2/img $1.25/txt, gpt-image-1: $2.50/img $1.25/txt,
+  // gpt-image-1-mini: $0.25/img $0.20/txt) but are not tracked here yet - add when the usage field is wired up.
+  'gpt-image-2':      { inputText: 5.00, inputImage:  8.00, outputImage: 30.00 },
+  'gpt-image-1.5':    { inputText: 5.00, inputImage:  8.00, outputImage: 32.00 },
+  'gpt-image-1':      { inputText: 5.00, inputImage: 10.00, outputImage: 40.00 },
+  'gpt-image-1-mini': { inputText: 2.00, inputImage:  2.50, outputImage:  8.00 },
   // Fixed pricing models handled separately in openAIImageModelsPricing()
   'dall-e-3': null,
   'dall-e-2': null,
@@ -235,8 +240,7 @@ function openAIImageModelsPrice(modelId: DalleModelId): undefined | { inputText:
  * TODO: update this when the OpenAI pricing changes.
  */
 export function openAIImageModelsPricing(modelId: DalleModelId, quality: DalleImageQuality, size: DalleSize): string {
-  // GPT Image family (gpt-image-1.5, gpt-image-1, gpt-image-1-mini)
-  if (modelId === 'gpt-image-1.5' || modelId === 'gpt-image-1' || modelId === 'gpt-image-1-mini') {
+  if (getImageModelFamily(modelId) === 'gpt-image') {
 
     // gpt-image-1-mini does not support high quality
     if (modelId === 'gpt-image-1-mini' && quality === 'high') quality = 'medium';
diff --git a/src/modules/t2i/dalle/store-module-dalle.ts b/src/modules/t2i/dalle/store-module-dalle.ts
index 1dd770513..8a3ef1a28 100644
--- a/src/modules/t2i/dalle/store-module-dalle.ts
+++ b/src/modules/t2i/dalle/store-module-dalle.ts
@@ -11,7 +11,8 @@ export const DALLE_DEFAULT_IMAGE_SIZE: DalleImageSize = '1024x1024'; // this wor
 export type DalleImageSize = DalleSizeGI | DalleSizeD3 | DalleSizeD2;
 
 // Note: 'chatgpt-image-latest' also exists with same pricing as gpt-image-1.5
-export type DalleModelId = 'gpt-image-1.5' | 'gpt-image-1' | 'gpt-image-1-mini' | 'dall-e-3' | 'dall-e-2';
+type GPTImageMoldelId = 'gpt-image-2' | 'gpt-image-1.5' | 'gpt-image-1' | 'gpt-image-1-mini';
+export type DalleModelId = GPTImageMoldelId | 'dall-e-3' | 'dall-e-2';
 export type DalleModelSelection = DalleModelId | null; // null = auto-select latest
 
 /**
@@ -22,7 +23,7 @@ export type DalleModelSelection = DalleModelId | null; // null = auto-select lat
 export function resolveDalleModelId(selection: DalleModelSelection): DalleModelId {
   // Auto-select latest model when null
   if (selection === null) {
-    return 'gpt-image-1.5'; // Current latest model
+    return 'gpt-image-2'; // Current latest image drawing model
   }
   return selection;
 }
@@ -40,7 +41,7 @@ export function resolveDalleModelId(selection: DalleModelSelection): DalleModelI
  * - Each family can have its own settings/pricing structure
  */
 export function getImageModelFamily(modelId: DalleModelId): 'gpt-image' | 'dall-e-3' | 'dall-e-2' {
-  if (modelId === 'gpt-image-1.5' || modelId === 'gpt-image-1' || modelId === 'gpt-image-1-mini')
+  if (modelId === 'gpt-image-2' || modelId === 'gpt-image-1.5' || modelId === 'gpt-image-1' || modelId === 'gpt-image-1-mini')
     return 'gpt-image';
   if (modelId === 'dall-e-3')
     return 'dall-e-3';