Today, Vantage is launching support for ChatGPT Enterprise costs, expanding the OpenAI integration beyond Codex and OpenAI Platform API costs to cover ChatGPT spend in a ChatGPT Enterprise workspace. Vantage now imports daily ChatGPT cost and token usage from OpenAI's Daily Usage Analytics API as a new ChatGPT service, broken out by user, model, and token type. Both ChatGPT and Codex costs now carry Managed AI Tags, so teams can report on them alongside their other AI providers.

ChatGPT Enterprise is OpenAI's edition of ChatGPT for organizations, giving employees access to OpenAI models for chat, research, data analysis, and coding with Codex, with admin controls and workspace analytics. Previously, Vantage's OpenAI integration imported OpenAI Platform API costs and Codex spend for OpenAI Enterprise customers. As a result, workspace totals in Vantage came in below what OpenAI reported, and teams had to check ChatGPT spend in the ChatGPT Admin console.

Now, with ChatGPT Enterprise cost support, customers can view their per-user costs incurred from ChatGPT alongside the rest of their AI and infrastructure spend. Customers with an existing Codex connection can keep it and add a separate workspace Admin key with Enterprise analytics API: Read. Vantage pairs the two connections for the same workspace and reports ChatGPT separately from Codex. Once connected, customers can group ChatGPT spend by user, model, or token type in Cost Reports, allocate it to teams with Virtual Tags, track it with Budgets and Anomaly Detection, and more.

Support for ChatGPT Enterprise costs is now available to all Vantage customers. To connect, go to the Integrations page in account settings and select OpenAI. You can learn more via the OpenAI section in Vantage's documentation.

Frequently Asked Questions

1. What is being launched today?

Vantage now imports ChatGPT Enterprise costs through the OpenAI integration. Daily ChatGPT cost and token usage from a ChatGPT Enterprise workspace appear in Vantage as a ChatGPT service, next to Codex costs from the same workspace and OpenAI Platform API costs. ChatGPT and Codex costs also carry Managed AI Tags.

2. Who is this for?

Organizations on ChatGPT Enterprise that want their full ChatGPT and Codex spend in Vantage, grouped by user and model, and reported next to the rest of their AI and cloud costs. FinOps teams can allocate ChatGPT spend to teams, and engineering and IT leaders can see which users and models drive it.

3. Who has access to this integration?

The OpenAI integration is available to all Vantage customers across all subscription tiers. Importing ChatGPT and Codex costs requires a ChatGPT Enterprise workspace.

4. How much does this integration cost?

There is no additional fee for using the integration. However, OpenAI costs will be included in quota tier enforcement. If your OpenAI costs push you over your current tier limit, you may be prompted to upgrade. See the Pricing page for details.

5. How do I set this up?

In OpenAI:

1. Confirm your organization is on ChatGPT Enterprise.
2. In the OpenAI Admin Console, go to Credentials, then Admin keys and enable Enterprise analytics API: Read (enterprise.analytics.usage.read), also called Unified Analytics.
3. Copy each key and the Workspace ID from Workspace details in the ChatGPT Admin console.

In Vantage:

1. Go to the Integrations page in account settings and select OpenAI.
2. On the Connect tab, click Add API Key.
3. Paste the Unified Analytics key, enter the Workspace ID, and enter a Connection Name. Click Connect Account.

If you have not yet also performed the Codex integration, it is recommended to perform that as well.

6. I already connected a workspace for Codex. What do I need to do?

Keep your existing Codex key and connection, and add a separate key with only Enterprise analytics API: Read (enterprise.analytics.usage.read). Use the same Workspace ID and Vantage account so Vantage can pair the keys and report ChatGPT separately without duplicating Codex costs. You do not need to replace your Codex key.

If your existing key already has both permissions, keep that connection instead of adding another key with overlapping permissions.

7. What permissions are required on the OpenAI side?

To import ChatGPT workspace analytics, provide a workspace Admin key with Enterprise analytics API: Read (enterprise.analytics.usage.read) and the Workspace ID from the ChatGPT Admin console. For separate Codex reporting, we recommend connecting a second key with only Codex analytics API: Read (codex.enterprise.analytics.read) for the same workspace and Vantage account.

A Unified Analytics-only key is accepted. Without a Codex connection, all returned workspace costs, including Codex, are grouped under ChatGPT, retaining model, user, and token-type attribution where OpenAI provides it. A Codex-only key imports only Codex costs. One key with both permissions is also supported.

If your organization restricts API access by IP address, add Vantage's IP addresses, listed in the documentation, to your OpenAI allowlist.

8. What permissions are required on the Vantage side?

You must have a Vantage Organization Owner or Integration Owner role to add or remove the OpenAI integration. See the Role-Based Access Control documentation for details.

9. What can Vantage access?

Vantage has read-only access to aggregated daily usage analytics for the connected workspace. Vantage cannot perform cost-incurring actions, change workspace settings, or run tasks in ChatGPT or Codex.

10. Does Vantage store prompts, conversations, or generated code?

No. Vantage imports only daily cost and usage metadata, such as token counts, models, and the user email and user ID associated with the usage. It does not import prompts, conversations, files, or generated code.

11. Which costs are included, and which are not?

Vantage imports the daily usage-based costs OpenAI returns for the connected ChatGPT Enterprise workspace. These include:

• Chat: Web, desktop, iOS, Android, and other chat clients (CHAT_WEB, CHAT_DESKTOP, CHAT_IOS, CHAT_ANDROID, and CHAT_OTHER), grouped under the ChatGPT service.
• Agents in Chat: Usage reported as AGENTS_CHATGPT, also grouped under the ChatGPT service.
• Codex and Work: With a Codex connection, reported separately by client, with services such as Codex CLI, Codex Web, Codex VS Code, Codex Desktop, Codex Work Desktop, Codex Work Mobile, and Codex Work Web. With only a Unified Analytics connection, this usage is included in ChatGPT instead.

Agents in Slack are not currently reported through the Unified API; these costs will not appear in Vantage.

12. What dimensions can I filter and group by?

• Billing Account: the ChatGPT workspace ID
• Account: the Connection Name you entered when connecting
• Service: ChatGPT, or the Codex client when Codex access is connected. With only Unified Analytics access, ChatGPT includes Codex usage.
• Category: the model. Spend that OpenAI doesn't report against a specific model appears as uncategorized.
• Subcategory: token type (uncached input, cached input, output)
• Resource: the user's email, or the user ID when no email is available
• Provider tags: openai:billing_source (workspace_analytics for ChatGPT rows, codex_analytics for Codex rows), openai:client_id, and openai:cost_attribution
• Managed AI Tags: vntg:ai:model, vntg:ai:model_provider, vntg:ai:token_type, vntg:ai:inference_speed, vntg:ai:user_email, and vntg:ai:user_id

13. How are dollar amounts calculated?

When both Unified Analytics and Codex access are connected for the same workspace, Vantage subtracts each user's Codex cost from their Unified Analytics model cost for the same day. The remainder is reported under ChatGPT; the Codex connection reports Codex costs separately. This works with two paired keys or one key with both permissions.

Without Codex access, Vantage reports the full Unified Analytics workspace costs under ChatGPT, retaining the reported models and users. Native dollar costs and dollar estimates for credit-priced usage are preserved; estimates are not invoice charges.

Dollar amounts by token type are derived. Vantage splits each daily model cost across uncached input, cached input, and output tokens in proportion to token counts, so the token-type amounts add up to the cost OpenAI reported.

14. Are token counts available?

Yes. Usage for ChatGPT and Codex costs is measured in tokens, so you can report on token counts alongside cost in Cost Reports.

15. What happens when OpenAI reports usage without a dollar price?

Vantage keeps the token usage and records the cost as $0, with "Dollar price unavailable" in the row description, so token counts stay complete.

16. What about spend that isn't tied to a specific user?

Codex spend that OpenAI reports at the workspace level but doesn't attribute to a user appears on a reconciliation row tagged openai:cost_attribution = unattributed, as it does today.

17. Does this change my existing Codex costs or tags?

Codex costs keep the same services, dimensions, and provider tags such as openai:billing_source and openai:client_id, and now also carry Managed AI Tags.

18. Will ChatGPT costs be double counted with Codex or OpenAI Platform costs?

Paired Codex-only and Unified Analytics-only keys for the same workspace and Vantage account do not duplicate Codex spend after their imports finish. Vantage uses Codex data to subtract Codex from the Unified connection's ChatGPT costs. A single key with both permissions uses the same separation. Vantage rejects additional connections with overlapping permissions for the same workspace.

If you first connect only Unified Analytics, ChatGPT includes Codex usage. Adding its complementary Codex key refreshes the Unified connection's recent history to separate that spend. Wait for both connections' imports and historical refreshes to finish before comparing totals.

ChatGPT Enterprise workspace usage is separate from OpenAI Platform API costs, so an OpenAI Platform organization Admin key can be connected alongside these workspace keys.

19. Can I connect multiple ChatGPT workspaces?

Yes. For each ChatGPT workspace, we recommend one Codex-only key and one Unified Analytics-only key, paired by the same Workspace ID within the same Vantage account. Add each key as a separate connection. You do not need one key per OpenAI Platform project. To track OpenAI Platform API costs, add an OpenAI Platform organization Admin key separately.

20. How often is data refreshed?

ChatGPT and Codex costs refresh once daily.

21. How much history is imported?

When you connect, Vantage imports up to the last 90 days of ChatGPT and Codex costs available from OpenAI.

22. Are taxes or discounts included?

Tax is not included. Amounts reflect the cost OpenAI reports in its workspace analytics.

23. Are Active Resources available?

No. ChatGPT and Codex costs are available in Cost Reports and related features, such as Virtual Tags, Budgets, and Anomaly Detection.

24. What happens if I remove the integration?

Removing a connection removes its associated costs from Vantage. If you remove the Codex connection while keeping its paired Unified Analytics connection, Vantage refreshes the remaining connection's recent history so ChatGPT includes the full workspace costs, including Codex. Wait for that refresh to finish before comparing totals.