Vertex AI features a growing list of foundation models that you can test, deploy, and customize for use in your AI-based applications. Foundation models are fine-tuned for specific use cases and offered at different price points. This page summarizes the models that are available in the various APIs and gives you guidance on which models to choose by use case.
To learn more about all AI models and APIs on Vertex AI, see Explore AI models and APIs.
Gemini models
The following table summarizes the models available in the Gemini API:
Model name | Description | Specifications |
---|---|---|
Gemini 1.5 Flash ( gemini-1.5-flash ) |
Multimodal model that is designed for high volume, cost-effective applications. Gemini 1.5 Flash delivers speed and efficiency to build fast, lower-cost applications that don't compromise on quality. | Max total tokens (input and output): 1M Max output tokens: 8,192 Max raw image size: 20 MB Max base64 encoded image size: 7 MB Max images per prompt: 3,000 Max video length: 1 hour Max videos per prompt: 10 Max audio length: approximately 8.4 hours Max audio per prompt: 1 Max PDF size: 30 MB Training data: Up to May 2024 |
Gemini 1.5 Pro ( gemini-1.5-pro ) |
Multimodal model that supports adding image, audio, video, and PDF files in text or chat prompts for a text or code response. Gemini 1.5 Pro supports long-context understanding with up to 1 million tokens. | Max total tokens (input and output): 1M Max output tokens: 8,192 Max images per prompt: 3,000 Max video length (frames only): approximately one hour Max video length (frame and audio): approximately 45 minutes Max videos per prompt: 10 Max audio length: approximately 8.4 hours Max audio per prompt: 1 Max PDF size: 30 MB Training data: Up to May 2024 |
Gemini 1.0 Pro ( gemini-1.0-pro ) |
The best performing model with features for a wide range of text-only
tasks.
Supports only text as input. Supports supervised tuning. |
Max total tokens (input and output): 32,760 Max output tokens: 8,192 Training data: Up to Feb 2023 |
Gemini 1.0 Pro Vision ( gemini-1.0-pro-vision ) |
The best performing image/video understanding model to handle a broad
range of applications.
Supports text, image, and video as inputs. |
Max total tokens (input and output): 16,384 Max output tokens: 2,048 Max images per prompt: 16 Max video length: 2 minutes Max videos per prompt: 1 Training data: Up to Feb 2023 |
Gemini 1.0 Ultra (GA with allow list) | Google's most capable text model, optimized for complex tasks,
including instruction, code, and reasoning.
Supports only text as input. |
Max tokens input: 8,192 Max tokens output: 2,048 |
Gemini 1.0 Ultra Vision (GA with allow list) |
Google's most capable multimodal vision model, optimized to support joint text, images, and video inputs. | Max tokens input: 8,192 Max tokens output: 2,048 |
Gemini language support
Gemini models support the following languages:
Arabic (ar), Bengali (bn), Bulgarian (bg),
Chinese simplified and traditional (zh), Croatian (hr),
Czech (cs), Danish (da), Dutch (nl), English (en), Estonian (et), Finnish (fi),
French (fr), German (de), Greek (el), Hebrew (iw), Hindi (hi), Hungarian (hu),
Indonesian (id), Italian (it), Japanese (ja), Korean (ko), Latvian (lv),
Lithuanian (lt), Norwegian (no), Polish (pl), Portuguese (pt), Romanian (ro),
Russian (ru), Serbian (sr), Slovak (sk), Slovenian (sl), Spanish (es),
Swahili (sw), Swedish (sv), Thai (th), Turkish (tr), Ukrainian (uk),
Vietnamese (vi).
Embeddings models
The following table summarizes the models available in the Embeddings API.
Model name | Description | Specifications |
---|---|---|
Embeddings for text ( ) |
Returns embeddings for English text inputs.
Supports supervised tuning of "text-embedding-gecko" models, English only. |
Max token input: 3,072 (textembedding-gecko@001), 2,048 (others). Embedding dimension: text-embedding-004 : <=768 Others: 768. |
Embeddings for text multilingual(textembedding-gecko-multilingual@001,
|
Returns embeddings for text inputs of over 100 languages
Supports supervised tuning of the text-multilingual-embedding-002 model. Supports 100 languages. |
Max token input: 2,048 Embedding dimension: text-multilingual-embedding-002 : <=768 Others: 768. |
Embeddings for multimodal(multimodalembedding) |
Returns embedding for text, image and video inputs, to compare content
across different models.
Converts text, image, and video into the same vector space. Video only supports 1408 dimensions. English only |
Max token input: 32, Max image size: 20 MB, Max video length: Two mins, Embedding dimension: 128, 256, 512, or 1408 for text+image input, 1408 for video input. |
Embeddings language support
Text multilingual embedding models support the following languages:
Afrikaans, Albanian, Amharic, Arabic, Armenian, Azerbaijani, Basque,
Belarusian, Bengali, Bulgarian, Burmese, Catalan, Cebuano, Chichewa, Chinese,
Corsican, Czech, Danish, Dutch, English, Esperanto, Estonian, Filipino,
Finnish, French, Galician, Georgian, German, Greek, Gujarati,
Haitian Creole, Hausa, Hawaiian, Hebrew, Hindi, Hmong, Hungarian,
Icelandic, Igbo, Indonesian, Irish, Italian, Japanese, Javanese, Kannada,
Kazakh, Khmer, Korean, Kurdish, Kyrgyz, Lao, Latin, Latvian, Lithuanian,
Luxembourgish, Macedonian, Malagasy, Malay, Malayalam, Maltese, Maori,
Marathi, Mongolian, Nepali, Norwegian, Pashto, Persian, Polish,
Portuguese, Punjabi, Romanian, Russian, Samoan, Scottish Gaelic,
Serbian, Shona, Sindhi, Sinhala, Slovak, Slovenian, Somali, Sotho,
Spanish, Sundanese, Swahili, Swedish, Tajik, Tamil, Telugu, Thai,
Turkish, Ukrainian, Urdu, Uzbek, Vietnamese, Welsh, West Frisian,
Xhosa, Yiddish, Yoruba, Zulu.
Imagen model
The following table summarizes the models available in the Imagen API:
Model name | Description | Specifications |
---|---|---|
Imagen 2 ( imagegeneration@006) |
This model supports image generation and editing to create high
quality images in seconds.
The editing feature supports object removal and insertion, outpainting, and product editing |
Max image output: four Aspect ratio (for generation): 1:1, 9:16, 16:9, 3:4, 4:3 Resolution: ~1500 pixels (varies by aspect ratio) |
Imagen language support
The Imagen model supports the following languages:
English, Chinese (simplified), Chinese (traditional), Hindi, Japanese, Korean,
Portuguese, and Spanish.
Code completion model
The following table summarizes the models available in the Codey APIs:
Model name | Description | Specifications |
---|---|---|
Codey for Code Completion ( code-gecko ) |
A model fine-tuned to suggest code completion based on the context in code that's written. | Maximum input tokens: 2048 Maximum output tokens: 64 |
MedLM models
The following table summarizes the models available in the MedLM API:
Model name | Description | Specifications |
---|---|---|
MedLM-medium (medlm-medium ) |
A HIPAA-compliant suite of medically tuned models and APIs powered
by Google Research. This model helps healthcare practitioners with medical question and answer tasks, and summarization tasks for healthcare and medical documents. Provides better throughput and includes more recent data than the medlm-large model. |
Max tokens (input + output): 32,768 Max output tokens: 8,192 |
MedLM-large (medlm-large ) |
This model helps healthcare practitioners with medical question and answer tasks, and summarization tasks for healthcare and medical documents. | Max input tokens: 8,192 Max output tokens: 1,024 |
Model versions and lifecycle
Each Generative AI on Vertex AI language model is available in a stable version and an auto-updated version. See the following topics to learn about how model versioning works with Gemini models. To learn about Imagen on Vertex AI model versions and their lifecycle, see Imagen on Vertex AI model versions and lifecycle.
If you tune a Gemini model, then the tuned model shares the same discontinuation date as the base model that you used in the tuning process. For more information, see Overview of model tuning for Gemini.
Gemini stable version
A stable version of a Gemini model does not change and continues to be
available until its discontinuation date. See the tables in
Available Gemini stable model versions on this
page to learn the discontinuation dates of Gemini models. If you use a stable
after its discontinuation date, you need to switch to a newer available
stable version. You can identify the version of a stable model by the
three-digit number that's appended to the model name. For example,
gemini-1.0-pro-001
is version number one of the stable release of the
Gemini 1.0 Pro model.
Google releases stable versions on a regular cadence. You can switch from one stable version to another as long as the other version is still available. When you move from one stable version to a different stable version, you need to run your tuning jobs again because there might be prompt, output, and other differences between the versions.
To use the stable version of a Gemini model, append the three digit version
number to the model with a hyphen (-
). For example, to specify the stable
gemini-1.0-pro
model that's version six, append -006
to the model's
name:
https://us-central1-aiplatform.googleapis.com/v1/projects/my_project/locations/us-central1/publishers/google/models/gemini-1.0-pro-006
Available Gemini stable model versions
The following stable model versions are available for generally available Gemini models:
Gemini 1.5 Flash model | Release date | Discontinuation date |
---|---|---|
gemini-1.5-flash-001 | May 24, 2024 | May 24, 2025 |
Gemini 1.5 Pro model | Release date | Discontinuation date |
---|---|---|
gemini-1.5-pro-001 | May 24, 2024 | May 24, 2025 |
Gemini 1.0 Pro Vision model | Release date | Discontinuation date |
---|---|---|
gemini-1.0-pro-vision-001 | February 15, 2024 | February 15, 2025 |
Gemini 1.0 Pro model | Release date | Discontinuation date |
---|---|---|
gemini-1.0-pro-001 | February 15, 2024 | February 15, 2025 |
gemini-1.0-pro-002 | April 9, 2024 | April 9, 2025 |
Gemini auto-updated version
The auto-updated version of a Gemini model points to the most recent stable version. When a new stable version is released, the auto-updated version points to the new version. This means that if you specify the auto-updated version of a Gemini model in your code, it could behave differently without notice when the next stable version is released. Because of this, use a auto-updated version with caution if you tune your model.
To use the auto-updated version of a model, don't append anything to the model name.
For example, the following uses the auto-updated version of the
gemini-1.0-pro-vision
model:
https://us-central1-aiplatform.googleapis.com/v1/projects/my_project/locations/us-central1/publishers/google/models/gemini-1.0-pro-vision
Gemini auto-updated models
The following table shows the available auto-updated Gemini model versions and the stable version each references.
Model name | Auto-updated name | Referenced stable version |
---|---|---|
Gemini 1.0 Pro Vision | gemini-1.0-pro-vision |
gemini-1.0-pro-vision-001 |
Gemini 1.0 Pro | gemini-1.0-pro |
gemini-1.0-pro-002 |
Gemini preview version
The preview version of a Gemini model is a model that's in preview and not generally available (GA). A preview version of a model contains functionality that's not in the most recent latest or auto-updated version of a model. Because a preview model version isn't stable, it's not recommended for use in production.
Each preview model includes its release date as part of the name of the model
that you use in your code. The name pattern used by a preview model is
model_name-preview-MMDD
. For example, gemini-1.5-pro-preview-0409
is the
first preview version of the
Gemini 1.5 Pro model and it was released on April 9. When
a new preview version of a model is released, the previous version is updated
to point to the new preview version and is available until its discontinuation
date.
Gemini preview models
The following table shows the available preview Gemini model versions and the preview version each references.
Model name | Preview name | Discontinuation date |
---|---|---|
Gemini 1.5 Flash (Preview) | gemini-1.5-flash-preview-0514 |
June 24, 2024 |
Gemini 1.5 Pro (Preview) | gemini-1.5-pro-preview-0514 |
June 24, 2024 |
Gemini 1.5 Pro (Preview) | gemini-1.5-pro-preview-0409 (points to and uses
gemini-1.5-pro-preview-0514 ) |
June 14, 2024 |
Code completion stable model versions
The following stable model versions are available for generally available Generative AI models:
code-gecko model | Release date | Discontinuation date |
---|---|---|
code-gecko@002 | December 6, 2023 | October 9, 2024 |
code-gecko@001 | June 29, 2023 | July 6, 2024 |
Embeddings stable model versions
The following stable model versions are available for generally available Generative AI models:
textembedding-gecko model | Release date | Discontinuation date |
---|---|---|
text-embedding-004 | May 14, 2024 | May 14, 2025 |
text-embedding-preview-0409 | April 9, 2024 | June 27, 2024 |
text-multilingual-embedding-002 | May 14, 2024 | May 14, 2025 |
text-multilingual-embedding-preview-0409 | April 9, 2024 | June 27, 2024 |
textembedding-gecko@003 | December 12, 2023 | December 12, 2024 |
textembedding-gecko-multilingual@001 | November 2, 2023 | December 12, 2024 |
textembedding-gecko@002(regressed, but still supported) | November 2, 2023 | October 9, 2024 |
textembedding-gecko@001 | June 7, 2023 | October 9, 2024 |
multimodalembedding@001 | February 12, 2024 | February 12, 2025 |
MedLM language support
The MedLM model supports the English language.
Explore all models in Model Garden
Model Garden is a platform that helps you discover, test, customize, and deploy Google proprietary and select OSS models and assets. To explore the generative AI models and APIs that are available on Vertex AI, go to Model Garden in the Google Cloud console.
To learn more about Model Garden, including available models and capabilities, see Explore AI models in Model Garden.
What's next
- Try a quickstart tutorial using Vertex AI Studio or the Vertex AI API.
- Learn how to test text prompts.
- Learn how to test chat prompts.
- Explore pretrained models in Model Garden.
- Learn how to tune a foundation model.
- Learn about responsible AI best practices and Vertex AI's safety filters.