[[["容易理解","easyToUnderstand","thumb-up"],["確實解決了我的問題","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["難以理解","hardToUnderstand","thumb-down"],["資訊或程式碼範例有誤","incorrectInformationOrSampleCode","thumb-down"],["缺少我需要的資訊/範例","missingTheInformationSamplesINeed","thumb-down"],["翻譯問題","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],["上次更新時間:2025-09-02 (世界標準時間)。"],[],[],null,["# Migrate to Gemini from AutoML text\n\nGemini is a family of generative AI models developed by Google that is\ndesigned for multimodal use cases. If you haven't used Gemini models on\nVertex AI before, see the [Generative AI\nintroduction](/vertex-ai/generative-ai/docs/learn/overview).\n\nKey advantages of Gemini include the following:\n\n- **Enhanced performance** : The latest large language models (LLMs), such as\n Gemini Flash 1.5, demonstrate better understanding across a range of\n natural language tasks than the AutoML text model. For\n more information, see the publicly available [technical\n report](https://arxiv.org/pdf/2403.05530) from the Gemini team.\n\n- **Flexibility**: Gemini allows for both prompting (quick adaptation)\n and fine-tuning (deeper customization), catering to different project needs.\n This flexibility allows for rapid prototyping, testing, and deployment using\n prompting, with the option to fine-tune the Gemini model weights for\n optimal performance on specific tasks. Vertex AI offers both\n console-based fine-tuning and SDK and API options for programmatic control.\n\n- **Multipurpose and multimodal capabilities**: Gemini offers the\n ability to process text, images, and other modalities. This approach enables consistent use of a single format and model across various tasks. This flexibility allows the process to be easily adapted for different applications, streamlining and accelerating development efforts.\n\nGemini supports most features available in AutoML text.\nHowever, there are\n[differences](/vertex-ai/docs/start/automl-gemini-comparison), and the client\nlibraries don't support client integration backward compatibility. In other\nwords, you must plan to migrate your resources to benefit from Gemini\nfeatures.\n\nIf you are planning a new project, you should build your code, job, dataset, or\nmodel with Gemini. This lets you take advantage of the new features and\nservice improvements as they become available.\n\nRecommended steps for migrating to Gemini\n-----------------------------------------\n\n| **Note:** AutoML Text is no longer supported.\n\nUse the following recommended steps to update your existing code,\njobs, datasets, and models from AutoML text to Gemini.\n\n1. Read about the major differences between Gemini\n and AutoML text at\n [Gemini for AutoML text\n users](/vertex-ai/docs/start/automl-gemini-comparison).\n\n2. Review any potential changes in pricing (see\n [Gemini migration pricing](#compare-legacy)).\n\n3. Take inventory of your Google Cloud projects, code, jobs,\n datasets, models, and users with access to AutoML text. Use\n this information to determine which resources to migrate\n and ensure that the correct users have access to the migrated\n resources.\n\n4. Review any [changes to IAM roles](#changes-to-iam-roles), and then\n update service accounts and authentication for your resources.\n\n5. Migrate your resources using either of these two methods:\n\n - [Prompting](/vertex-ai/generative-ai/docs/models/gemini-tuning#gemini-prompts)\n\n - [Fine-tuning](/vertex-ai/generative-ai/docs/models/gemini-tuning#fine-tuning-gemini)\n\n6. View the [locations](/vertex-ai/generative-ai/docs/learn/locations)\n available for Gemini.\n\n7. [Identify usage of\n AutoML text APIs](#identifying-api-usage) to help determine\n which of your applications use them\n and to identify the method calls that you want to migrate.\n\n8. Update your applications and workflows to use Gemini.\n\n9. Plan your request quota monitoring. See [Quotas and\n limits](/vertex-ai/generative-ai/docs/quotas).\n\nGemini migration pricing\n------------------------\n\nMigration is free. After migration, legacy resources are still available to use\nin AutoML text until the service shuts down in June 2025. To avoid\nunnecessary costs, shut down or delete legacy resources after you have verified\nthat your objects have migrated successfully.\n\nGemini pricing compared to AutoML text pricing\n----------------------------------------------\n\nGemini pricing is generally cheaper compared to equivalent tasks in\nAutoML text. Gemini pricing is determined by whether you're\nusing the model for prompt engineering only, fine-tuning only, or a combination\nof both. For more information, you can compare [AutoML text\npricing](/vertex-ai/pricing#automl_models) with [Gemini\npricing](/vertex-ai/generative-ai/pricing).\n\nFor entity extraction models, consider that the model serving output may be higher as the output is the complete structured data.\n\nIdentify usage of AutoML text APIs\n----------------------------------\n\nYou can determine which of your applications use\nAutoML APIs, as well as\nwhich methods they are using. Use this information to help determine whether\nthese API calls need to be migrated to\nGemini:\n\n- For each of your projects, go to the [APIs \\&\n Services Dashboard](https://console.cloud.google.com/apis/dashboard)\n to see a list of\n which products' API the project uses. To learn more, see\n [Monitoring API usage](/apis/docs/monitoring).\n\n- If enabled, you can check the audit logs created by\n [AutoML text](/vertex-ai/docs/general/audit-logging)\n as part of [Cloud Audit Logs](/logging/docs/audit).\n\n- To see usage of specific AutoML text methods,\n go to the [AutoML text Metrics\n page](https://console.cloud.google.com/apis/api/aiplatform.googleapis.com/metrics).\n\nManage changes to IAM roles and permissions\n-------------------------------------------\n\nVertex AI provides the following Identity and Access Management (IAM) roles:\n\n- `aiplatform.admin`\n- `aiplatform.user`\n- `aiplatform.viewer`\n\nUsing Vertex AI datasets is no longer required. Data for fine-tuning in\nGemini can only be stored in Cloud Storage.\n\nFor more information on IAM roles, see [Access\ncontrol](/vertex-ai/docs/general/access-control).\n\nWhat's next\n-----------\n\n- Read [Introduction to tuning](/vertex-ai/docs/start/migrating-applications)\n in Gemini.\n\n- See [Gemini for AutoML text users](/vertex-ai/docs/start/automl-gemini-comparison) for a comparison of Gemini and AutoML text."]]