Skip to main content
Vantage integrates with a Together AI organization using an API key and imports daily rated costs from Together’s Billing Usage API. Costs are broken down by product, such as Dedicated Inference and Serverless Inference, and by the GPU type, model, project, and endpoint on each line item. Usage data is available for Dedicated Inference, measured in GPU-hours, and for Serverless Inference, measured in tokens.
Vantage imports the rated cost Together returns on each line item. Credits, discounts, and tax are not applied, so costs in Vantage can differ from your final Together AI invoice. Days with no line items are not imported. Daily costs run through yesterday, UTC.

​
Together AI Permissions

Vantage connects with a Together AI project API key and checks it by calling the Billing Usage API. That API returns usage for the key’s entire organization, not only the project the key belongs to. Project admins and members can create keys. See Together’s API key documentation for how keys are scoped. Together does not offer a billing-only API key. A project key can also run inference and spend credits. Vantage only reads billing usage. It does not create endpoints, run inference or fine-tuning, or store prompts or model outputs. Create a dedicated key for Vantage. Together shows a new key only once.

​
Connect Your Together AI Account

​
Prerequisites

  • You must have a Vantage Organization Owner or Integration Owner role to add or remove this integration. See the Role-Based Access Control documentation for details.
  • You need access to a Together AI project where you can create an API key.
  • At this time, the Billing Usage API is enabled per organization. Contact Together support to request access. Calls from an organization without access return a 404.
  • Create a free Vantage account, then follow the steps below to integrate Together AI costs.
If your organization restricts access to the provider by IP address, add the following Vantage IP addresses to your provider’s allowlist:
  • 54.87.66.45
  • 3.95.43.133
  • 54.162.3.72
  • 44.199.143.63
  • 3.218.103.23

​
Create the Connection

1
Sign in to Together AI and open the project where you want to create the key.
2
Open the project’s API keys settings.
3
Click Create key, enter a name, and select Create. You can set an expiration date. If the key expires, Vantage can no longer refresh costs until you connect a new key.
4
Copy the API key. Together shows new keys only once.
5
From the top navigation in Vantage, click Settings.
6
On the left navigation, select Integrations and then Together AI.
7
Open the Connect tab and click Add API Key.
8
Complete the integration form:
  • API Key: Paste the Together AI API key.
  • Description: Enter a name that identifies this connection in Vantage. This label is shown on the Together AI integration page. It is not the Organization value in Cost Reports.
9
Click Connect Account.
Vantage checks that the key can read billing usage before it imports costs. If the key is invalid, Vantage rejects it. If Together has not enabled the Billing Usage API for that organization, the connection fails and no costs are imported. After clicking Connect Account, the integration status changes to Importing. Vantage then imports up to six months of historical Together AI costs. See the Integration Status documentation for details on these statuses. Connect one key per Together AI organization. The Billing Usage API returns that organization’s costs, including every project, so a second key for the same organization imports the same costs again. To connect another organization, click Add Another API Key and repeat these steps. Removing the connection in Vantage does not revoke the API key in Together.
As soon as costs are processed, they are available on your All Resources Cost Report. If you remove the Together AI integration, the costs for that connection are removed from the Vantage console.

​
Next Steps - Manage Workspace Access

Once the import is complete and the integration status changes to Stable, you can select which workspaces this integration is associated with. See the Workspaces documentation for information.

​
Data Refresh

See the provider data refresh documentation for information on when data for each provider refreshes in Vantage. Together AI data refreshes once daily. During the first five days of each month, Vantage also refreshes the previous month so late-arriving charges are captured.

​
Together AI Reporting Dimensions

On Together AI Cost Reports, you can filter and group across several dimensions:
  • Organization (the Together organization ID)
  • Account (the Together project ID, when Together includes one)
  • Service (the Together product, such as Dedicated Inference, Serverless Inference, GPU Clusters, Fine-Tuning, Code Sandbox, Managed Storage, or Provisioned Throughput)
  • Category (the GPU type on dedicated rows, such as H100-80GB; the model name on serverless rows; the product name when Together does not send pricing dimensions)
  • Subcategory (the usage class, such as On-demand GPU-hours, Reserved GPU-hours, Input tokens (cached), Input tokens (uncached), or Output tokens. Batch token usage is prefixed with Batch ·)
  • Resource (the Together endpoint name, when the line item has one)
  • Charge Type (Usage)
  • Tags, including together:capacity_type (BASE or BURST on dedicated GPU rows) and Managed AI Tags
Vantage applies Managed AI Tags so you can compare Together AI with other AI providers. Serverless rows can include vntg:ai:model, vntg:ai:token_type (input, output, or cache_read), and vntg:ai:api_key_id. vntg:ai:model_provider is written when the model name identifies a known model builder. vntg:ai:endpoint is the Together endpoint name when the line item includes one. Dedicated GPU rows typically have vntg:ai:endpoint and do not have a model or token type.