针对导致错误的每个 API 请求，例如包含错误格式 JSON 的用户事件请求或价格为负的目录项导入请求，Vertex AI Search for Commerce 会将错误记录到 Google Cloud Observability 中。
对于目录项未包含在导入目录中的每个预测请求，Vertex AI Search for Commerce 也会记录相应错误。
[[["易于理解","easyToUnderstand","thumb-up"],["解决了我的问题","solvedMyProblem","thumb-up"],["其他","otherUp","thumb-up"]],[["很难理解","hardToUnderstand","thumb-down"],["信息或示例代码不正确","incorrectInformationOrSampleCode","thumb-down"],["没有我需要的信息/示例","missingTheInformationSamplesINeed","thumb-down"],["翻译问题","translationIssue","thumb-down"],["其他","otherDown","thumb-down"]],[],[],[],null,["# Pricing\n\nVertex AI Search for commerce pricing\n=====================================\n\nPrices are listed in US Dollars (USD).\nIf you pay in a currency other than USD, the prices listed in your currency on\n[Cloud Platform SKUs](https://cloud.google.com/skus/)\napply.\n\nSearch charges\n--------------\n\nSearch enables you to provide high quality product results that\nare customizable for your business needs. You can leverage Google's query and\ncontextual understanding to improve product discovery across your website and\nmobile applications.\n\nThe only search operations that incur charges are requesting\nsearch or browse results by calling the [Search](/retail/docs/reference/rest/v2/projects.locations.catalogs.placements/search#google.cloud.retail.v2.SearchService.Search) method. There's no charge for importing or managing user events or\ncatalog information. There is also no charge for [using the pretrained Recommendations LLM](/retail/docs/models-pretrained-llm).\n\nSearch and browse queries are charged at $2.50 per\n1000 requests.\n\n### Example\n\nThis example illustrates how search queries are charged.\n\nIn this example, a customer's application made 15 million keyword search\nqueries and 10 million browse queries in one month. Here is how we would\ncalculate the total cost to the customer:\n\n- Search queries = 15 million\n- Browse queries = 10 million\n- Total queries for the month = 15 million + 10 million = 25 million\n- Vertex AI Search for commerce search pricing = $2.50/1000 queries\n\nTotal cost to the customer = 25 million queries x $2.50/1000 queries = $62,500\n\nRecommendations charges\n-----------------------\n\n**Free trial:** You can try recommendations with $600 free credits.\nFree credits are automatically granted when you sign up, and expire six months\nafter signing up. These credits are granted to your billing account, and are not\naffected by the number of projects connected to your billing account. For\nexample, if you have three projects linked to your billing account, your billing\naccount still receives $600 of free credits. These credits are typically\nsufficient to train a model and test its performance in production through a\ntwo-week A/B test. See [Implement Vertex AI Search for commerce](/retail/docs/overview).\n\nThere's no charge for importing or managing user events or catalog information.\nThe only recommendations operations that incur charges are training,\ntuning, or requesting predictions by calling the\n[predict](/retail/docs/reference/rest/v2/projects.locations.catalogs.placements/predict) method.\n\nTraining (per node per hour) costs are charged on a daily basis if your model\nis actively training or if you have submitted a request to resume training. Once\nyou pause or delete the model, you will no longer be charged. See the\ndocumentation for [managing training](/retail/docs/manage-models#pause).\n\nTuning (per node per hour) costs for active models are charged after the tune\ncompletes successfully. You will only be charged for an incomplete tune if you\npause or delete a model during an ongoing tune. In this case, you will then be\ncharged for the node hours that were consumed before the model tuning stopped.\nSee the documentation for [managing tuning](/retail/docs/manage-models#tune).\n\n### Examples\n\n#### Example A\n\nThis example illustrates how each tier of pricing for monthly prediction\nrequests is applied.\n\nIn this example, a large retailer's application made 1,000,000,000 prediction\nrequests in this particular month. It trains three models which, by default,\nautomatically retrain once per day. This amounts to about 500 node hours of\nmodel training a month. By default, recommendations models are tuned\nquarterly; in this example, the model tuning accrued about 300 node hours per\ntuning, which on a monthly basis would be 100 node hours.\n\nTo calculate the cost for this month, we'll first find the cost of prediction\nrequests. Pricing is calculated in 1000-request blocks, and cost is tiered by\nnumber of monthly prediction requests.\n\n- First 20,000,000 predictions = 20,000,000 predictions / 1000 \\* $0.27 = $5,400\n- Next 280,000,000 predictions = 280,000,000 predictions / 1000 \\* $0.18 = $50,400\n- Next 700,000,000 predictions = 700,000,000 predictions / 1000 \\* $0.10 = $70,000\n\nNext, let's calculate the cost of training and tuning.\n\n- Training charge = 500 node hours \\* $2.50 = $1,250\n- Tuning charge = 100 node hours \\* $2.50 = $250\n\nThe total cost of predictions, training, and tuning in the given month is\n$127,300.\n\n#### Example B\n\nThis example illustrates a lower volume use case.\n\nIn this example, a retailer makes 10,000,000 prediction requests a month and\ntrains a single model per day, which, by default, automatically retrains once\nper day. This amounts to about 150 node hours of model training per month. The\nmodel's quarterly tuning accrued about 90 node hours per tuning; to find the\ncost per month, we'll use the monthly average, 30 node hours.\n\nLet's calculate the price for one month of usage. Because this retailer's number\nof prediction requests this month doesn't exceed 20,000,000, requests will all\nbe charged at the first tier of pricing,\n$0.27 per 1000 requests.\n\n- 10,000,000 predictions = 10,000,000 predictions / 1000 \\* $0.27 = $2,700\n\nTo calculate the cost of training and tuning:\n\n- Training charge = 150 node hours \\* $2.50 = $375\n- Tuning charge = 30 node hours \\* $2.50 = $75\n\nThe total cost of predictions, training, and tuning in the given month is\n$3,150.\n\nGoogle Cloud Observability charges\n----------------------------------\n\nVertex AI Search for commerce [logs an error to Google Cloud Observability](/retail/docs/error-reporting)\nfor each API request that results in an error, such as a user event request that\ncontains malformed JSON, or a catalog item import request with a negative price.\nVertex AI Search for commerce also logs an error for every prediction request with a\ncatalog item that is not in the imported catalog.\n\nGoogle Cloud Observability charges by the GiB of logs stored. (Logs are retained\nfor one month.) The first 50 GiB of logs per month per project is free. After\nthat, Google Cloud Observability charges $0.50 per GiB of logs. The size of the\nlogging data depends on the size of your JSON payload, but a GiB would\nbe approximately 200,000 recommendations errors.\n\nFor more information, see the\n[Google Cloud Observability pricing page](/stackdriver/pricing).\n\nWhat's next\n-----------\n\n- Read the [Vertex AI Search for commerce documentation](/retail/docs).\n- Try the [Pricing calculator](/products/calculator).\n- Learn about [Vertex AI Search for commerce solutions and use cases](/architecture?text=Vertex AI Search for commerce).\n\n#### Request a custom quote\n\nWith Google Cloud's pay-as-you-go pricing, you only pay for the services you use. Connect with our sales team to get a custom quote for your organization.\n[Contact sales](/contact?direct=true)"]]
