Skip to content

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

feat(google-gemini): clearer forms, aspect ratio cards and a shorter Create Video step (#15826)

Talal Jaberde693f195f8e
diff --git a/packages/pieces/community/google-gemini/package.json b/packages/pieces/community/google-gemini/package.json
index c262a7425e0..60ae1f0413a 100644
--- a/packages/pieces/community/google-gemini/package.json
+++ b/packages/pieces/community/google-gemini/package.json
@@ -1,6 +1,6 @@
{
"name": "@activepieces/piece-google-gemini",
- "version": "0.3.0",
+ "version": "0.4.0",
"main": "./dist/src/index.js",
"types": "./dist/src/index.d.ts",
"dependencies": {
diff --git a/packages/pieces/community/google-gemini/src/index.ts b/packages/pieces/community/google-gemini/src/index.ts
index ecb901bca3e..4a69c7e729d 100644
--- a/packages/pieces/community/google-gemini/src/index.ts
+++ b/packages/pieces/community/google-gemini/src/index.ts
@@ -16,8 +16,8 @@ import { googleGeminiAuth } from './lib/auth';
export const googleGemini = createPiece({
displayName: 'Google Gemini',
auth: googleGeminiAuth,
- description: 'Use the new Gemini models from Google',
- minimumSupportedRelease: '0.87.0',
+ description: 'Generate text, images, video and speech with Google Gemini.',
+ minimumSupportedRelease: '0.88.2',
logoUrl: 'https://cdn.activepieces.com/pieces/google-gemini.png',
categories: [PieceCategory.ARTIFICIAL_INTELLIGENCE],
authors: ["pfernandez98","kishanprmr","MoShizzle","AbdulTheActivePiecer","abuaboud"],
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/chat-gemini.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/chat-gemini.action.ts
index 36bcf4504b4..6f433a3ff77 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/chat-gemini.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/chat-gemini.action.ts
@@ -16,28 +16,28 @@ export const chatGemini = createAction({
auth: googleGeminiAuth,
name: 'chat_gemini',
classification: 'READ',
- displayName: 'Chat Gemini',
- description: 'Chat with Google Gemini',
+ displayName: 'Chat with Gemini',
+ description: 'Chat with Gemini, optionally remembering earlier messages.',
aiMetadata: { description: 'Sends a prompt as one turn of a Gemini chat and returns the reply, with two modes: supply a memory key (max 128 characters) to load and persist the conversation history in project storage so later runs remember earlier turns, or leave it empty for a stateless single turn. Pick it over generate_content when the exchange spans multiple runs and must retain context; generate_content is the better choice for one-shot generation or when a built-in tool such as Google Search or File Search is needed. Not idempotent: each call produces a new reply and, when a memory key is set, rewrites the stored history.', idempotent: false },
props: {
+ prompt: Property.LongText({
+ displayName: 'Prompt',
+ required: true,
+ description: 'Your next message in the conversation.',
+ }),
model: Property.Dropdown({
displayName: 'Model',
required: true,
- description: 'The model which will generate the completion',
+ description: 'Gemini model that writes the response.',
refreshers: [],
auth: googleGeminiAuth,
defaultValue: defaultLLM,
options: async ({ auth }) => getGeminiModelOptions({ auth }),
}),
- prompt: Property.LongText({
- displayName: 'Prompt',
- required: true,
- description: 'The prompt to generate content from.',
- }),
memoryKey: Property.ShortText({
displayName: 'Memory Key',
description:
- 'A memory key that will keep the chat history. Keep it empty to leave Gemini without memory of previous messages.',
+ 'Reuse a key to continue its conversation. Empty: no memory.',
required: false,
}),
},
@@ -47,7 +47,7 @@ export const chatGemini = createAction({
memoryKey: z.optional(z.string().check(z.maxLength(128))),
});
- const { model, prompt, memoryKey } = propsValue;
+ const { model, prompt, memoryKey } = propsValue;
const genAI = new GoogleGenerativeAI(auth.secret_text);
const geminiModel = genAI.getGenerativeModel({ model });
let history: Content[] = [];
@@ -65,15 +65,15 @@ export const chatGemini = createAction({
const result = await chat.sendMessage(prompt);
const responseText = result.response.text();
+ const returnedHistory = memoryKey ? await chat.getHistory() : history;
if (memoryKey) {
- const updatedHistory = await chat.getHistory();
- await store.put(memoryKey, updatedHistory, StoreScope.PROJECT);
+ await store.put(memoryKey, returnedHistory, StoreScope.PROJECT);
}
return {
response: responseText,
- history: history,
+ history: returnedHistory,
};
},
});
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/count-tokens.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/count-tokens.action.ts
index 3f4bef1c88e..6c0a61f78a2 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/count-tokens.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/count-tokens.action.ts
@@ -25,6 +25,7 @@ export const countTokensAction = createAction({
model: Property.Dropdown({
displayName: 'Model',
required: true,
+ description: 'Each model counts tokens differently.',
auth: googleGeminiAuth,
refreshers: [],
defaultValue: defaultLLM,
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/create-video.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/create-video.action.ts
index 74807f3fa3f..75e7bd51cf7 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/create-video.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/create-video.action.ts
@@ -15,47 +15,35 @@ export const createVideoAction = createAction({
description: 'Generate a video from a text prompt using Google Veo models.',
aiMetadata: { description: 'Generates a short video with a Google Veo model from a text prompt, optionally seeded by a start image (image-to-video) or by both a start and end image (interpolation between those frames), polling the long-running generation job for up to 10 minutes before returning the finished MP4 as a file. Use it whenever the required output is video; text goes through generate_content and speech through text-to-speech. Note the model-dependent limits: 1080p needs Veo 3.0 or newer, 4K needs Veo 3.1, and both require an 8-second duration. Not idempotent: each call renders a new video.', idempotent: false },
props: {
+ prompt: Property.LongText({
+ displayName: 'Prompt',
+ description: 'Describe the scene, motion, style and lighting you want.',
+ required: true,
+ }),
model: Property.Dropdown({
displayName: 'Model',
+ description: 'Veo model that creates the video.',
required: true,
auth: googleGeminiAuth,
refreshers: [],
defaultValue: 'veo-3.1-generate-preview',
options: async ({ auth }) => getGeminiVideoModelOptions({ auth }),
}),
- prompt: Property.LongText({
- displayName: 'Prompt',
- description:
- 'Describe the video you want to generate. Be specific about the scene, motion, style, and lighting for best results.',
- required: true,
- }),
- image: Property.File({
- displayName: 'Start Image',
- description:
- 'Optional image to use as the first frame of the video (image-to-video).',
- required: false,
- }),
- lastFrame: Property.File({
- displayName: 'End Image',
- description:
- 'Optional image to use as the last frame. Use together with Start Image to generate an interpolation video.',
- required: false,
- }),
aspectRatio: Property.StaticDropdown({
displayName: 'Aspect Ratio',
required: true,
defaultValue: '16:9',
+ display: 'cards',
options: {
- disabled: false,
options: [
- { label: '16:9 (Landscape)', value: '16:9' },
- { label: '9:16 (Portrait)', value: '9:16' },
+ { label: 'Landscape', value: '16:9', description: '16:9' },
+ { label: 'Portrait', value: '9:16', description: '9:16' },
],
},
}),
durationSeconds: Property.Dropdown({
- displayName: 'Duration (seconds)',
- description: 'Must be 8 when using 1080p, 4K, or reference images.',
+ displayName: 'Duration',
+ description: 'Video length in seconds. 1080p and 4K need 8 seconds.',
required: false,
auth: googleGeminiAuth,
refreshers: ['model'],
@@ -86,7 +74,7 @@ export const createVideoAction = createAction({
resolution: Property.Dropdown({
displayName: 'Resolution',
description:
- '1080p is available on Veo 3.0+ only. 4K is only available on Veo 3.1 models. 1080p and 4K require 8-second duration.',
+ '4K needs Veo 3.1. 1080p and 4K need an 8-second video.',
required: false,
auth: googleGeminiAuth,
refreshers: ['model'],
@@ -105,18 +93,29 @@ export const createVideoAction = createAction({
};
},
}),
+ image: Property.File({
+ displayName: 'Start Image',
+ description: 'Image to use as the first frame of the video.',
+ required: false,
+ advanced: true,
+ }),
+ lastFrame: Property.File({
+ displayName: 'End Image',
+ description: 'Image to use as the last frame. Needs a Start Image.',
+ required: false,
+ advanced: true,
+ }),
personGeneration: Property.StaticDropdown({
- displayName: 'Person Generation',
- description:
- 'Controls whether people or faces can appear in the generated video.',
+ displayName: 'People in Video',
+ description: 'Whether people can appear in the video.',
required: false,
+ advanced: true,
defaultValue: 'allow_adult',
options: {
- disabled: false,
options: [
- { label: 'Allow Adults Only (default)', value: 'allow_adult' },
- { label: 'Allow All Ages (requires allowlist)', value: 'allow_all' },
- { label: "Don't Allow People", value: 'dont_allow' },
+ { label: 'Adults Only', value: 'allow_adult' },
+ { label: 'All Ages (requires allowlist)', value: 'allow_all' },
+ { label: 'No People', value: 'dont_allow' },
],
},
}),
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/generate-content-from-image.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/generate-content-from-image.action.ts
index 36c4e6fe3af..5dd4b5ee880 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/generate-content-from-image.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/generate-content-from-image.action.ts
@@ -11,7 +11,7 @@ import { generateContentFromImageActionOutputSchema } from '../output-schemas';
export const generateContentFromImageAction = createAction({
audience: 'both',
description:
- 'Generate content using Google Gemini using the "gemini-pro-vision" model',
+ 'Ask Gemini about an image and get a text response.',
aiMetadata: { description: 'Sends an image together with a prompt to a vision-capable Gemini model and returns the generated text describing or answering questions about that image, covering captioning, text extraction, and visual question answering. This is the only Gemini action here that accepts image input, so pick it over generate_content whenever a picture is part of the question; the image is required and is sent inline, so keep it small. Not idempotent: each call produces a fresh completion.', idempotent: false },
displayName: 'Generate Content from Image',
name: 'generate_content_from_image',
@@ -21,18 +21,18 @@ export const generateContentFromImageAction = createAction({
prompt: Property.LongText({
displayName: 'Prompt',
required: true,
- description: 'The prompt to generate content from.',
+ description: 'What you want to know about the image.',
}),
image: Property.File({
displayName: 'Image',
required: true,
- description: 'The image to generate content from.'
+ description: 'The image Gemini looks at.'
}),
model: Property.Dropdown({
displayName: 'Model',
auth: googleGeminiAuth,
required: true,
- description: 'The model which will generate the completion',
+ description: 'Gemini model that writes the response.',
refreshers: [],
defaultValue: defaultLLM,
options: async ({ auth }) =>
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/generate-content-with-file-search.ts b/packages/pieces/community/google-gemini/src/lib/actions/generate-content-with-file-search.ts
index 0d5321c321f..c11a5b4651e 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/generate-content-with-file-search.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/generate-content-with-file-search.ts
@@ -7,7 +7,7 @@ import { generateContentWithFilesearchActionOutputSchema } from '../output-schem
export const generateContentWithFileSearchAction = createAction({
audience: 'both',
- description: 'Generate content with file search functionality.',
+ description: 'Upload a file and have Gemini answer a prompt from it.',
aiMetadata: { description: 'Uploads one file to a newly created Gemini File Search store, waits for indexing to finish, then answers the prompt grounded on that document. Use it for one-shot question answering over a document supplied at run time; prefer generate_content when no file grounding is needed or when a different built-in tool such as Google Search or URL Context fits better. Both a file and a store display name are required. Not idempotent: every call creates another file search store and a fresh completion.', idempotent: false },
displayName: 'Generate Content with File Search',
name: 'generate_content_with_filesearch',
@@ -17,24 +17,26 @@ export const generateContentWithFileSearchAction = createAction({
prompt: Property.LongText({
displayName: 'Prompt',
required: true,
- description: 'The prompt to generate content from.',
+ description: 'Question or instruction Gemini answers from the file.',
}),
model: Property.Dropdown({
displayName: 'Model',
auth: googleGeminiAuth,
required: true,
- description: 'The model which will generate the completion',
+ description: 'Gemini model that writes the response.',
refreshers: [],
defaultValue: defaultLLM,
options: async ({ auth }) => getGeminiModelOptions({ auth }),
}),
- fileStoreName: Property.ShortText({
- displayName: 'File Store Name',
- required: true,
- }),
file: Property.File({
displayName: 'File',
required: true,
+ description: 'File Gemini searches to answer the prompt.',
+ }),
+ fileStoreName: Property.ShortText({
+ displayName: 'File Store Name',
+ required: true,
+ description: 'Name for the search store this step creates on each run.',
}),
},
outputSchema: generateContentWithFilesearchActionOutputSchema,
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/generate-content.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/generate-content.action.ts
index 6602281397b..7060b36e933 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/generate-content.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/generate-content.action.ts
@@ -9,7 +9,7 @@ import { generateContentActionOutputSchema } from '../output-schemas';
export const generateContentAction = createAction({
audience: 'both',
- description: 'Generate content using Google Gemini using the "gemini-pro" model',
+ description: 'Send a prompt to Gemini and get a text response.',
aiMetadata: { description: 'Runs a single stateless prompt through a Gemini text model and returns the generated text, optionally grounded by one built-in tool: Google Search for live web results, URL Context to fetch pages named in the prompt, Google Maps scoped to a latitude/longitude, or File Search over an uploaded file. Use this as the default Gemini text-generation call; prefer chat_gemini when the exchange needs conversation memory, generate_content_from_image when the input includes an image, and create_video or text-to-speech for non-text output. Not idempotent: each call produces a fresh completion, and the File Search mode additionally creates a new file search store.', idempotent: false },
displayName: 'Generate Content',
name: 'generate_content',
@@ -19,23 +19,22 @@ export const generateContentAction = createAction({
prompt: Property.LongText({
displayName: 'Prompt',
required: true,
- description: 'The prompt to generate content from.',
+ description: 'What you want Gemini to write or answer.',
}),
model: Property.Dropdown({
displayName: 'Model',
required: true,
- description: 'The model which will generate the completion',
+ description: 'Gemini model that writes the response.',
refreshers: [],
defaultValue: defaultLLM,
auth: googleGeminiAuth,
options: async ({ auth }) => getGeminiModelOptions({ auth }),
}),
toolType: Property.StaticDropdown({
- displayName: 'Tool Type',
- description: 'Select built-in tool to use with Gemini model.',
+ displayName: 'Built-in Tool',
+ description: 'Lets Gemini search Google, read links, use Maps or search a file.',
required: false,
options: {
- disabled: false,
options: [
{ label: 'Google Search', value: 'google-search' },
{ label: 'File Search', value: 'file-search' },
@@ -45,7 +44,7 @@ export const generateContentAction = createAction({
},
}),
toolProperties: Property.DynamicProperties({
- displayName: 'Tool Config',
+ displayName: 'Tool Settings',
auth: googleGeminiAuth,
refreshers: ['toolType'],
required: false,
@@ -60,11 +59,12 @@ export const generateContentAction = createAction({
file: Property.File({
displayName: 'File',
required: true,
- description: 'File to use for search tool.',
+ description: 'File Gemini searches to answer the prompt.',
}),
fileStoreName: Property.ShortText({
displayName: 'File Store Name',
required: true,
+ description: 'Name for the search store this step creates on each run.',
}),
};
break;
@@ -73,7 +73,7 @@ export const generateContentAction = createAction({
mrkdown: Property.MarkDown({
variant: MarkdownVariant.INFO,
value:
- 'To use URL context tool, include one or more URLs directly in the prompt. Gemini will fetch and use the page content as additional context.',
+ 'Put the links in the prompt. Gemini reads those pages before it answers.',
}),
};
break;
@@ -82,7 +82,7 @@ export const generateContentAction = createAction({
mrkdown: Property.MarkDown({
variant: MarkdownVariant.INFO,
value:
- 'To use Google Search tool, include a clear search query in the prompt. Gemini will use Google Search to retrieve up-to-date web information.',
+ 'Gemini searches Google when the prompt needs up-to-date information.',
}),
};
break;
@@ -91,12 +91,12 @@ export const generateContentAction = createAction({
latitude: Property.Number({
displayName: 'Latitude',
required: true,
- description: 'Provide the relevant location latitude.',
+ description: 'Center of the Maps search, in decimal degrees from -90 to 90.',
}),
longitude: Property.Number({
displayName: 'Longitude',
required: true,
- description: 'Provide the relevant location longitude.',
+ description: 'Center of the Maps search, in decimal degrees from -180 to 180.',
}),
};
break;
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/generate-embeddings.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/generate-embeddings.action.ts
index 919a871caec..259972fd386 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/generate-embeddings.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/generate-embeddings.action.ts
@@ -10,7 +10,7 @@ export const generateEmbeddingsAction = createAction({
classification: 'READ',
auth: googleGeminiAuth,
displayName: 'Generate Embeddings',
- description: 'Converts text into a numerical embedding vector using a Gemini embedding model.',
+ description: 'Turns text into a list of numbers for search and similarity.',
aiMetadata: {
description:
'Converts input text into a numerical embedding vector for semantic search, similarity comparison, clustering, or classification. Set Task Type to the intended use (e.g. retrieval_document when indexing content, retrieval_query when searching it) for a better-tuned vector. Idempotent: the same text and settings always produce the same vector.',
@@ -24,6 +24,7 @@ export const generateEmbeddingsAction = createAction({
}),
model: Property.Dropdown({
displayName: 'Model',
+ description: 'Embedding model that turns the text into numbers.',
required: true,
auth: googleGeminiAuth,
refreshers: [],
@@ -47,14 +48,16 @@ export const generateEmbeddingsAction = createAction({
},
}),
title: Property.ShortText({
- displayName: 'Title',
- description: 'Only used when Task Type is Retrieval Document; improves embedding quality for document search.',
+ displayName: 'Document Title',
+ description: 'Only used when Task Type is Retrieval Document.',
required: false,
+ advanced: true,
}),
outputDimensionality: Property.Number({
- displayName: 'Output Dimensionality',
- description: 'Truncate the embedding to this many dimensions. Only supported on newer models.',
+ displayName: 'Vector Dimensions',
+ description: 'Shorten the vector to this many numbers. Empty: full length.',
required: false,
+ advanced: true,
}),
},
outputSchema: generateEmbeddingsActionOutputSchema,
@@ -68,7 +71,7 @@ export const generateEmbeddingsAction = createAction({
contents: text,
config: {
taskType,
- title,
+ ...(taskType === 'RETRIEVAL_DOCUMENT' ? { title } : {}),
outputDimensionality,
},
});
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/generate-image.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/generate-image.action.ts
index bb4e2194f7f..f241de44f28 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/generate-image.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/generate-image.action.ts
@@ -25,6 +25,7 @@ export const generateImageAction = createAction({
}),
model: Property.Dropdown({
displayName: 'Model',
+ description: 'Gemini image model that creates the picture.',
required: true,
auth: googleGeminiAuth,
refreshers: [],
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/list-models.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/list-models.action.ts
index c3799f33ad7..3cb42abed84 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/list-models.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/list-models.action.ts
@@ -9,7 +9,7 @@ export const listModelsAction = createAction({
classification: 'SEARCH',
auth: googleGeminiAuth,
displayName: 'List Models',
- description: 'Lists the Gemini and Veo models available to this API key.',
+ description: 'Lists the models this API key can use, with limits where known.',
aiMetadata: {
description:
'Lists every Gemini/Veo model available to this API key, with its token limits and supported generation methods. Use this to discover valid model names before calling generate_content, generate_image, create_video, or generate_embeddings, or to check a model\'s input/output token limits. Safe to retry: read-only.',
diff --git a/packages/pieces/community/google-gemini/src/lib/actions/text-to-speech.action.ts b/packages/pieces/community/google-gemini/src/lib/actions/text-to-speech.action.ts
index 3d671d3c70c..a2fecf0c908 100644
--- a/packages/pieces/community/google-gemini/src/lib/actions/text-to-speech.action.ts
+++ b/packages/pieces/community/google-gemini/src/lib/actions/text-to-speech.action.ts
@@ -12,20 +12,22 @@ export const textToSpeechAction = createAction({
classification: 'READ',
auth: googleGeminiAuth,
displayName: 'Text to Speech',
- description: 'Converts text to audio file.',
+ description: 'Turn text into a spoken WAV audio file.',
aiMetadata: { description: 'Synthesizes spoken audio from input text using a Gemini TTS model and one of roughly thirty prebuilt named voices, returning a WAV file. This is the audio-output sibling of generate_content: use it when the flow needs speech instead of text, and create_video when it needs video. Requires a TTS-capable model, which the model dropdown restricts to. Not idempotent: each call renders a new audio file.', idempotent: false },
props: {
+ text: Property.LongText({
+ displayName: 'Text',
+ description: 'The words Gemini reads aloud.',
+ required: true,
+ }),
model: Property.Dropdown({
displayName: 'Model',
+ description: 'Gemini speech model that reads the text aloud.',
required: true,
auth: googleGeminiAuth,
refreshers: [],
options: async ({ auth }) => getGeminiTtsModelOptions({ auth }),
}),
- text: Property.LongText({
- displayName: 'Input Text',
- required: true,
- }),
voice: Property.StaticDropdown({
displayName: 'Voice',
required: true,
@@ -102,7 +104,7 @@ export const textToSpeechAction = createAction({
});
} catch (error) {
- console.error('Error in generate content from image:', error);
+ console.error('Error in text to speech:', error);
throw error;
}
},
diff --git a/packages/pieces/community/google-gemini/src/lib/auth.ts b/packages/pieces/community/google-gemini/src/lib/auth.ts
index c015fa8cb78..6e0acc81b07 100644
--- a/packages/pieces/community/google-gemini/src/lib/auth.ts
+++ b/packages/pieces/community/google-gemini/src/lib/auth.ts
@@ -2,10 +2,11 @@ import { PieceAuth } from '@activepieces/pieces-framework';
import { httpClient, HttpMethod } from '@activepieces/pieces-common';
const markdownDescription = `
-Follow these instructions to get your API Key:
-1. Visit the following website: https://makersuite.google.com/app/apikey
-2. Once on the website, locate and click on the option to obtain your API Key.
-Please note this piece uses a API in the beta phase that may change at any time.
+To get your API key:
+1. Open [Google AI Studio](https://aistudio.google.com/app/apikey).
+2. Click **Create API key** and copy the key.
+
+This piece uses Gemini's beta API, which Google may change at any time.
`;
export const googleGeminiAuth = PieceAuth.SecretText({
diff --git a/packages/pieces/community/google-gemini/src/lib/common/common.ts b/packages/pieces/community/google-gemini/src/lib/common/common.ts
index ff4833c00be..6753fcba729 100644
--- a/packages/pieces/community/google-gemini/src/lib/common/common.ts
+++ b/packages/pieces/community/google-gemini/src/lib/common/common.ts
@@ -32,6 +32,7 @@ const isVeoModel = (model: GeminiModel): boolean =>
const isImageModel = (model: GeminiModel): boolean =>
model.name.toLowerCase().includes('image') &&
+ !model.name.toLowerCase().includes('imagen') &&
!isTtsModel(model) &&
!isVeoModel(model);