@chrischall/gemini-mcp 2.3.4 → 2.4.0

package/.claude-plugin/marketplace.json

@@ -7,7 +7,7 @@
   },
   "metadata": {
     "description": "MCP server for Google Gemini image generation and editing — generate and edit images via the Gemini API",
-    "version": "2.3.4"
+    "version": "2.4.0"
   },
   "plugins": [
     {
@@ -15,7 +15,7 @@
       "displayName": "Gemini Images",
       "source": "./",
       "description": "Generate and edit images with Google Gemini image models (Nano Banana / Nano Banana Pro) via natural language prompts",
-      "version": "2.3.4",
+      "version": "2.4.0",
       "author": {
         "name": "Chris Hall"
       },

package/.claude-plugin/plugin.json

@@ -1,7 +1,7 @@
 {
   "name": "gemini-mcp",
   "displayName": "Gemini Images",
-  "version": "2.3.4",
+  "version": "2.4.0",
   "description": "MCP server for Google Gemini image generation and editing — generate and edit images via the Gemini API",
   "author": {
     "name": "Chris Hall",

package/README.md

@@ -13,7 +13,7 @@ Developed and maintained by AI (Claude Code).
 | Variable | Required | Description |
 |---|---|---|
 | `GEMINI_API_KEY` | Yes | Your Google Gemini API key ([aistudio.google.com/apikey](https://aistudio.google.com/apikey)) |
-| `GEMINI_IMAGE_MODEL` | No | Override the default image model (default: `gemini-3.1-flash-image`) |
+| `GEMINI_IMAGE_MODEL` | No | Override the default image model (default: `gemini-nano-banana-2.1`) |
 | `GEMINI_OUTPUT_DIR` | No | Default directory for generated images (default: ~/Downloads/gemini-mcp). When set, a per-call `output_dir` must be inside it. |
 | `GEMINI_INPUT_DIR` | No | Directory to resolve bare input-image filenames against (so `images: ["foo.jpg"]` works) |
 | `GEMINI_UPLOAD_DIR` | No | Restrict local files streamed to the Gemini Files API (`gemini_upload_file` `path`, `video_path`) to these directories — one or more, separated by `:` (`;` on Windows); `~` allowed. A path outside is refused before any upload. Unset: no restriction |

package/dist/bundle.js

@@ -23439,10 +23439,10 @@ var require_v4 = __commonJS({
   }
 });
 
-// node_modules/@modelcontextprotocol/core/dist/auth-BNDyLTqp.mjs
+// node_modules/@modelcontextprotocol/server/node_modules/@modelcontextprotocol/core/dist/auth-BNDyLTqp.mjs
 var z, LATEST_PROTOCOL_VERSION, SUPPORTED_PROTOCOL_VERSIONS, RELATED_TASK_META_KEY, PROTOCOL_VERSION_META_KEY, CLIENT_INFO_META_KEY, SERVER_INFO_META_KEY, CLIENT_CAPABILITIES_META_KEY, SUBSCRIPTION_ID_META_KEY, LOG_LEVEL_META_KEY, JSONRPC_VERSION, JSONValueSchema, JSONObjectSchema, JSONArraySchema, ProgressTokenSchema, CursorSchema, TaskMetadataSchema, RelatedTaskMetadataSchema, RequestMetaSchema, BaseRequestParamsSchema, TaskAugmentedRequestParamsSchema, RequestSchema, NotificationsParamsSchema, NotificationSchema, ResultMetaObjectSchema, ResultSchema, RequestIdSchema, JSONRPCRequestSchema, JSONRPCNotificationSchema, JSONRPCResultResponseSchema, JSONRPCErrorResponseSchema, JSONRPCMessageSchema, JSONRPCResponseSchema, EmptyResultSchema, CancelledNotificationParamsSchema, CancelledNotificationSchema, IconSchema, IconsSchema, BaseMetadataSchema, ImplementationSchema, FormElicitationCapabilitySchema, ElicitationCapabilitySchema, ClientTasksCapabilitySchema, ServerTasksCapabilitySchema, ClientCapabilitiesSchema, InitializeRequestParamsSchema, InitializeRequestSchema, ServerCapabilitiesSchema, InitializeResultSchema, InitializedNotificationSchema, DiscoverRequestSchema, DiscoverResultSchema, PingRequestSchema, ProgressSchema, ProgressNotificationParamsSchema, ProgressNotificationSchema, PaginatedRequestParamsSchema, PaginatedRequestSchema, PaginatedResultSchema, ResourceContentsSchema, TextResourceContentsSchema, Base64Schema, BlobResourceContentsSchema, RoleSchema, AnnotationsSchema, ResourceSchema, ResourceTemplateSchema, ListResourcesRequestSchema, ListResourcesResultSchema, ListResourceTemplatesRequestSchema, ListResourceTemplatesResultSchema, ResourceRequestParamsSchema, ReadResourceRequestParamsSchema, ReadResourceRequestSchema, ReadResourceResultSchema, ResourceListChangedNotificationSchema, SubscribeRequestParamsSchema, SubscribeRequestSchema, UnsubscribeRequestParamsSchema, UnsubscribeRequestSchema, SubscriptionFilterSchema, SubscriptionsListenRequestParamsSchema, SubscriptionsListenRequestSchema, SubscriptionsAcknowledgedNotificationParamsSchema, SubscriptionsAcknowledgedNotificationSchema, SubscriptionsListenResultMetaSchema, SubscriptionsListenResultSchema, ResourceUpdatedNotificationParamsSchema, ResourceUpdatedNotificationSchema, PromptArgumentSchema, PromptSchema, ListPromptsRequestSchema, ListPromptsResultSchema, GetPromptRequestParamsSchema, GetPromptRequestSchema, TextContentSchema, ImageContentSchema, AudioContentSchema, ToolUseContentSchema, EmbeddedResourceSchema, ResourceLinkSchema, ContentBlockSchema, PromptMessageSchema, GetPromptResultSchema, PromptListChangedNotificationSchema, ToolAnnotationsSchema, ToolExecutionSchema, ToolSchema, ListToolsRequestSchema, ListToolsResultSchema, CallToolResultSchema, CompatibilityCallToolResultSchema, CallToolRequestParamsSchema, CallToolRequestSchema, ToolListChangedNotificationSchema, ListChangedOptionsBaseSchema, LoggingLevelSchema, SetLevelRequestParamsSchema, SetLevelRequestSchema, LoggingMessageNotificationParamsSchema, LoggingMessageNotificationSchema, ModelHintSchema, ModelPreferencesSchema, ToolChoiceSchema, ToolResultContentSchema, SamplingContentSchema, SamplingMessageContentBlockSchema, SamplingMessageSchema, CreateMessageRequestParamsSchema, CreateMessageRequestSchema, CreateMessageResultSchema, CreateMessageResultWithToolsSchema, BooleanSchemaSchema, StringSchemaSchema, NumberSchemaSchema, UntitledSingleSelectEnumSchemaSchema, TitledSingleSelectEnumSchemaSchema, LegacyTitledEnumSchemaSchema, SingleSelectEnumSchemaSchema, UntitledMultiSelectEnumSchemaSchema, TitledMultiSelectEnumSchemaSchema, MultiSelectEnumSchemaSchema, EnumSchemaSchema, PrimitiveSchemaDefinitionSchema, ElicitRequestFormParamsSchema, ElicitRequestURLParamsSchema, ElicitRequestParamsSchema, ElicitRequestSchema, ElicitationCompleteNotificationParamsSchema, ElicitationCompleteNotificationSchema, ElicitResultSchema, ResourceTemplateReferenceSchema, PromptReferenceSchema, CompleteRequestParamsSchema, CompleteRequestSchema, CompleteResultSchema, RootSchema, ListRootsRequestSchema, ListRootsResultSchema, RootsListChangedNotificationSchema, TaskCreationParamsSchema, TaskStatusSchema, TaskSchema, CreateTaskResultSchema, TaskStatusNotificationParamsSchema, TaskStatusNotificationSchema, GetTaskRequestSchema, GetTaskResultSchema, GetTaskPayloadRequestSchema, GetTaskPayloadResultSchema, ListTasksRequestSchema, ListTasksResultSchema, CancelTaskRequestSchema, CancelTaskResultSchema, ClientRequestSchema, ClientNotificationSchema, ClientResultSchema, ServerRequestSchema, ServerNotificationSchema, ServerResultSchema, SafeUrlSchema, OAuthProtectedResourceMetadataSchema, OAuthMetadataSchema, OpenIdProviderMetadataSchema, OpenIdProviderDiscoveryMetadataSchema, OAuthTokensSchema, IdJagTokenExchangeResponseSchema, OAuthErrorResponseSchema, OptionalSafeUrlSchema, OAuthClientMetadataSchema, OAuthClientInformationSchema, OAuthClientInformationFullSchema, OAuthClientRegistrationErrorSchema, OAuthTokenRevocationRequestSchema;
 var init_auth_BNDyLTqp = __esm({
-  "node_modules/@modelcontextprotocol/core/dist/auth-BNDyLTqp.mjs"() {
+  "node_modules/@modelcontextprotocol/server/node_modules/@modelcontextprotocol/core/dist/auth-BNDyLTqp.mjs"() {
     z = __toESM(require_v4(), 1);
     LATEST_PROTOCOL_VERSION = "2025-11-25";
     SUPPORTED_PROTOCOL_VERSIONS = [
@@ -24386,9 +24386,9 @@ var init_auth_BNDyLTqp = __esm({
   }
 });
 
-// node_modules/@modelcontextprotocol/core/dist/internal.mjs
+// node_modules/@modelcontextprotocol/server/node_modules/@modelcontextprotocol/core/dist/internal.mjs
 var init_internal = __esm({
-  "node_modules/@modelcontextprotocol/core/dist/internal.mjs"() {
+  "node_modules/@modelcontextprotocol/server/node_modules/@modelcontextprotocol/core/dist/internal.mjs"() {
     init_auth_BNDyLTqp();
   }
 });
@@ -61227,7 +61227,7 @@ async function loadStdioDotenv() {
 }
 
 // src/version.ts
-var VERSION = "2.3.4";
+var VERSION = "2.4.0";
 
 // src/errors.ts
 init_dist2();
@@ -61511,14 +61511,14 @@ init_zod();
 init_dist2();
 
 // src/models.ts
-var DEFAULT_IMAGE_MODEL = "gemini-3.1-flash-image";
+var DEFAULT_IMAGE_MODEL = "gemini-nano-banana-2.1";
 var DEFAULT_VIDEO_MODEL = "gemini-omni-1.1-flash";
 var DEFAULT_MUSIC_MODEL = "lyria-3-clip-preview";
 function resolveModel(perCall, envOverride) {
   return perCall?.trim() || envOverride?.trim() || DEFAULT_IMAGE_MODEL;
 }
 function filterImageModels(raw) {
-  return raw.filter((m) => /image/i.test(m.name ?? "") && !/imagen/i.test(m.name ?? "")).map((m) => ({
+  return raw.filter((m) => /image|nano-banana/i.test(m.name ?? "") && !/imagen/i.test(m.name ?? "")).map((m) => ({
     id: (m.name ?? "").replace(/^models\//, ""),
     displayName: m.displayName ?? "",
     description: m.description ?? ""
@@ -62399,7 +62399,7 @@ function resolveAspectRatio(opts, supported) {
 var orientationSchema = external_exports.enum(ORIENTATIONS).optional().describe(
   "Output shape in plain terms: landscape (16:9), portrait (9:16) or square (1:1). For any other proportion \u2014 3:2, 4:3, 4:5, 21:9 \u2014 use aspect_ratio, which wins if both are given."
 );
-var MODEL_CHOICE_GUIDE = "gemini-3.1-flash-image (Nano Banana 2, the default) is the generalist: fast, 4K, reliable text, strong multi-reference consistency. gemini-3-pro-image (Pro) is for the hardest work \u2014 best world knowledge, brand precision. gemini-3.1-flash-lite-image (Lite) is cheapest: 1K only, no search grounding.";
+var MODEL_CHOICE_GUIDE = "gemini-nano-banana-2.1 (Nano Banana 2.1, the default) is the generalist: fast, 4K, reliable text, strong multi-reference consistency. gemini-3-pro-image (Pro) is for the hardest work \u2014 best world knowledge, brand precision. gemini-3.1-flash-lite-image (Lite) is cheapest: 1K only, no search grounding.";
 var timeoutMsSchema = external_exports.number().int().positive().optional().describe(
   "Upstream timeout in ms for this call (default $GEMINI_TIMEOUT_MS, else 60000 \u2014 120000 at 4K, which runs past 60s)"
 );
@@ -62502,7 +62502,7 @@ var sharedImageSchema = {
   model: external_exports.string().regex(/^[\w.-]+$/, 'must be a bare model id (letters, digits, ".", "_", "-")').optional().describe(`Model id override (default: server default; see gemini_list_models). ${MODEL_CHOICE_GUIDE}`),
   aspect_ratio: external_exports.enum(ASPECT_RATIOS).optional().describe("Exact output aspect ratio. For a plain landscape/portrait/square request, `orientation` is the shorthand; this wins if both are given."),
   orientation: orientationSchema,
-  image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution (512 = 0.5K, Flash-only)"),
+  image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution (512 = 0.5K, gemini-3.1-flash-image only)"),
   output_dir: external_exports.string().optional().describe("Directory to write images to (default: $GEMINI_OUTPUT_DIR or ~/Downloads/gemini-mcp)"),
   inline: external_exports.boolean().optional().describe("Return the image as an inline image block you can SEE, instead of a path (stdio) or link (hosted). The default costs nothing to carry and hands back a reference you can reuse; use this when you need to check the result yourself. On the hosted connector gemini_view_media does the same for an image you already have"),
   seed: external_exports.number().int().optional().describe("Seed for reproducible generation; random if omitted"),
@@ -63336,6 +63336,10 @@ var PRICED_AT = "2026-09-09";
 var RATE_CARD = Object.freeze({
   "gemini-3-pro-image": { input: 2, text_output: 12, image_output: 120 },
   "gemini-3.1-flash-image": { input: 0.5, text_output: 3, image_output: 60 },
+  // Nano Banana 2.1 (GA 2026-10-06; rates read 2026-10-14, after PRICED_AT):
+  // a pricier input and text rate than 2 but half the image rate, so a 1K
+  // image is $0.0336 rather than $0.067.
+  "gemini-nano-banana-2.1": { input: 1.5, text_output: 7.5, image_output: 30 },
   "gemini-3.1-flash-lite-image": { input: 0.25, text_output: 1.5, image_output: 30 },
   // Google publishes 2.5-flash-image's output as $0.039 PER IMAGE at 1290
   // tokens, not as a per-1M rate. 0.039 / 1290 * 1e6 = 30.23 — a back-derived
@@ -65345,7 +65349,7 @@ function registerInteractTools(server, client2) {
         model: external_exports.string().optional().describe(`Model id override (default: server default; see gemini_list_models). ${MODEL_CHOICE_GUIDE}`),
         aspect_ratio: external_exports.enum(ASPECT_RATIOS).optional().describe("Exact output aspect ratio. For a plain landscape/portrait/square request, `orientation` is the shorthand; this wins if both are given."),
         orientation: orientationSchema,
-        image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution (512 = 0.5K, Flash-only)"),
+        image_size: external_exports.enum(IMAGE_SIZES).optional().describe("Output resolution (512 = 0.5K, gemini-3.1-flash-image only)"),
         thinking_level: external_exports.enum(["minimal", "high"]).optional().describe("Reasoning depth; higher can help complex/structural edits"),
         filename: external_exports.string().optional().describe("Base filename for the output image (extension stripped; default: slugified input)"),
         output_dir: external_exports.string().optional().describe("Directory to write images to (default: $GEMINI_OUTPUT_DIR or ~/Downloads/gemini-mcp)"),

package/dist/models.js

@@ -1,10 +1,13 @@
 /**
- * The generalist workhorse (Nano Banana 2) — overridable per-call and via
- * GEMINI_IMAGE_MODEL. The premium gemini-3-pro-image is a deliberate opt-in:
- * it costs more and needs a funded account, so callers reach for it via the
- * `model` param when the task warrants it (see MODEL_CHOICE_GUIDE).
+ * The generalist workhorse (Nano Banana 2.1) — overridable per-call and via
+ * GEMINI_IMAGE_MODEL. Replaced gemini-3.1-flash-image (Nano Banana 2), which
+ * Google deprecated 2026-10-06 with no shutdown date yet; 2.1 is also half the
+ * per-image price ($0.0336 vs $0.067 at 1K). The premium gemini-3-pro-image
+ * is a deliberate opt-in: it costs more and needs a funded account, so callers
+ * reach for it via the `model` param when the task warrants it (see
+ * MODEL_CHOICE_GUIDE).
  */
-export const DEFAULT_IMAGE_MODEL = 'gemini-3.1-flash-image';
+export const DEFAULT_IMAGE_MODEL = 'gemini-nano-banana-2.1';
 /**
  * Default video model (omni, Interactions API) — overridable per-call.
  *
@@ -28,10 +31,12 @@ export function resolveModel(perCall, envOverride) {
  * Keep only Gemini image-generation models (the Nano Banana family) and strip
  * the `models/` prefix. Excludes `imagen-*` — those contain the substring
  * "image" but use a different `:predict` API this server doesn't implement.
+ * `nano-banana` is matched too: `gemini-nano-banana-2.1`, the default, has no
+ * "image" in its id and would otherwise be invisible to gemini_list_models.
  */
 export function filterImageModels(raw) {
     return raw
-        .filter((m) => /image/i.test(m.name ?? '') && !/imagen/i.test(m.name ?? ''))
+        .filter((m) => /image|nano-banana/i.test(m.name ?? '') && !/imagen/i.test(m.name ?? ''))
         .map((m) => ({
         id: (m.name ?? '').replace(/^models\//, ''),
         displayName: m.displayName ?? '',

package/dist/pricing.js

@@ -57,6 +57,10 @@ export const PRICED_AT = '2026-09-09';
 export const RATE_CARD = Object.freeze({
     'gemini-3-pro-image': { input: 2.0, text_output: 12.0, image_output: 120.0 },
     'gemini-3.1-flash-image': { input: 0.5, text_output: 3.0, image_output: 60.0 },
+    // Nano Banana 2.1 (GA 2026-10-06; rates read 2026-10-14, after PRICED_AT):
+    // a pricier input and text rate than 2 but half the image rate, so a 1K
+    // image is $0.0336 rather than $0.067.
+    'gemini-nano-banana-2.1': { input: 1.5, text_output: 7.5, image_output: 30.0 },
     'gemini-3.1-flash-lite-image': { input: 0.25, text_output: 1.5, image_output: 30.0 },
     // Google publishes 2.5-flash-image's output as $0.039 PER IMAGE at 1290
     // tokens, not as a per-1M rate. 0.039 / 1290 * 1e6 = 30.23 — a back-derived

package/dist/tools/interact.js

@@ -121,7 +121,7 @@ export function registerInteractTools(server, client) {
             image_size: z
                 .enum(IMAGE_SIZES)
                 .optional()
-                .describe('Output resolution (512 = 0.5K, Flash-only)'),
+                .describe('Output resolution (512 = 0.5K, gemini-3.1-flash-image only)'),
             thinking_level: z
                 .enum(['minimal', 'high'])
                 .optional()

package/dist/tools/shared.js

@@ -14,7 +14,10 @@ import { buildZip } from '../zip.js';
 const DEFAULT_SINK = createDiskSink();
 /** Supported output aspect ratios (Gemini image API). */
 export const ASPECT_RATIOS = ['1:1', '2:3', '3:2', '3:4', '4:3', '4:5', '5:4', '9:16', '16:9', '21:9', '1:4', '4:1', '1:8', '8:1'];
-/** Supported output resolutions. `512` is Flash-only (0.5K); `1K`/`2K`/`4K` work on all image models. */
+/**
+ * Supported output resolutions. `512` (0.5K) is gemini-3.1-flash-image only — Nano Banana 2.1, the
+ * default, documents 1K/2K/4K — while `1K`/`2K`/`4K` work on all image models.
+ */
 export const IMAGE_SIZES = ['512', '1K', '2K', '4K'];
 /**
  * The words people actually use for shape, as a shorthand over
@@ -77,7 +80,7 @@ export const orientationSchema = z
  * When to pick which Nano Banana model — shared by every tool's `model` param
  * so the guidance can't drift between them.
  */
-export const MODEL_CHOICE_GUIDE = 'gemini-3.1-flash-image (Nano Banana 2, the default) is the generalist: fast, 4K, reliable text, strong ' +
+export const MODEL_CHOICE_GUIDE = 'gemini-nano-banana-2.1 (Nano Banana 2.1, the default) is the generalist: fast, 4K, reliable text, strong ' +
     'multi-reference consistency. gemini-3-pro-image (Pro) is for the hardest work — best world knowledge, brand ' +
     'precision. gemini-3.1-flash-lite-image (Lite) is cheapest: 1K only, no search grounding.';
 /**
@@ -299,7 +302,7 @@ export const sharedImageSchema = {
         .optional()
         .describe('Exact output aspect ratio. For a plain landscape/portrait/square request, `orientation` is the shorthand; this wins if both are given.'),
     orientation: orientationSchema,
-    image_size: z.enum(IMAGE_SIZES).optional().describe('Output resolution (512 = 0.5K, Flash-only)'),
+    image_size: z.enum(IMAGE_SIZES).optional().describe('Output resolution (512 = 0.5K, gemini-3.1-flash-image only)'),
     output_dir: z.string().optional().describe('Directory to write images to (default: $GEMINI_OUTPUT_DIR or ~/Downloads/gemini-mcp)'),
     inline: z.boolean().optional().describe('Return the image as an inline image block you can SEE, instead of a path (stdio) or link (hosted). The default costs nothing to carry and hands back a reference you can reuse; use this when you need to check the result yourself. On the hosted connector gemini_view_media does the same for an image you already have'),
     seed: z.number().int().optional().describe('Seed for reproducible generation; random if omitted'),

package/dist/version.js

@@ -1,2 +1,2 @@
 /** Single source of the server version. release-please bumps the literal below. */
-export const VERSION = '2.3.4'; // x-release-please-version
+export const VERSION = '2.4.0'; // x-release-please-version

package/mint.yaml

@@ -13,7 +13,7 @@ env:
   - name: GEMINI_IMAGE_MODEL
     required: false
     help: >-
-      Override the default image model (default: gemini-3.1-flash-image)
+      Override the default image model (default: gemini-nano-banana-2.1)
   - name: GEMINI_OUTPUT_DIR
     required: false
     help: >-

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "@chrischall/gemini-mcp",
-  "version": "2.3.4",
+  "version": "2.4.0",
   "mcpName": "io.github.chrischall/gemini-mcp",
   "description": "Gemini image-generation MCP server for Claude — developed and maintained by AI (Claude Code)",
   "author": "Claude Code (AI) <https://www.anthropic.com/claude>",

package/server.json

@@ -6,12 +6,12 @@
     "url": "https://github.com/chrischall/gemini-mcp",
     "source": "github"
   },
-  "version": "2.3.4",
+  "version": "2.4.0",
   "packages": [
     {
       "registryType": "npm",
       "identifier": "@chrischall/gemini-mcp",
-      "version": "2.3.4",
+      "version": "2.4.0",
       "transport": {
         "type": "stdio"
       },
@@ -25,7 +25,7 @@
         },
         {
           "name": "GEMINI_IMAGE_MODEL",
-          "description": "Override the default image model (default: gemini-3.1-flash-image)",
+          "description": "Override the default image model (default: gemini-nano-banana-2.1)",
           "isRequired": false,
           "format": "string",
           "isSecret": false

package/skills/gemini-mcp/SKILL.md

@@ -69,7 +69,7 @@ Note: Image generation requires a billing-enabled Google Cloud project.
 | Variable | Required | Description |
 |---|---|---|
 | `GEMINI_API_KEY` | Yes | Your Google Gemini API key |
-| `GEMINI_IMAGE_MODEL` | No | Override the default image model (default: `gemini-3.1-flash-image`) |
+| `GEMINI_IMAGE_MODEL` | No | Override the default image model (default: `gemini-nano-banana-2.1`) |
 | `GEMINI_OUTPUT_DIR` | No | Default directory for saved images (default: ~/Downloads/gemini-mcp) |
 | `GEMINI_INPUT_DIR` | No | Directory to resolve bare input-image filenames against (e.g. point at Cowork's `uploads/` folder so `images: ["house.jpg"]` works) |
 
@@ -84,7 +84,8 @@ Note: Image generation requires a billing-enabled Google Cloud project.
 
 | Model | When to use | Reference-image caps (of 14 max) |
 |-------|-------------|----------------------------------|
-| `gemini-3.1-flash-image` (Nano Banana 2) | The versatile generalist workhorse for all tasks — balances speed with state-of-the-art 4K generation, world knowledge, and reliable text rendering; excels at multi-reference-image processing and consistency. Only model with video input + image_search grounding | 10 objects + 4 characters + 3 style refs |
+| `gemini-nano-banana-2.1` (Nano Banana 2.1, **default**) | The versatile generalist workhorse for all tasks — an update to Nano Banana 2 with better visual quality, prompt adherence, character consistency and text rendering; 1K/2K/4K, $0.0336 per 1K image | up to 10 objects + 4 characters |
+| `gemini-3.1-flash-image` (Nano Banana 2, **deprecated** 2026-10-06, no shutdown date yet) | The previous generalist workhorse — balances speed with state-of-the-art 4K generation, world knowledge, and reliable text rendering; excels at multi-reference-image processing and consistency. Only model with video input + image_search grounding | 10 objects + 4 characters + 3 style refs |
 | `gemini-3-pro-image` (Nano Banana Pro) | The premium choice for the most complex visual tasks — highest world knowledge, advanced localization, accurate brand consistency, precision creative control | 6 objects + 5 characters |
 | `gemini-3.1-flash-lite-image` (Nano Banana 2 Lite) | The fastest/cheapest for simple tasks — 1K output only, no Google Search grounding | 14 objects (no character consistency) |