Today, Vantage announces Managed AI Tags, a new tagging schema that standardizes AI cost data across providers, giving customers a consistent way to understand, analyze, and manage their AI spend. Customers can filter, group, and allocate spend by dimensions like model, token type, and user, while avoiding the need to build and maintain custom Virtual Tags.

As teams adopt AI tools from Cursor, Anthropic, OpenAI, AWS Bedrock, Azure OpenAI, and Google Cloud, FinOps and Engineering leaders need a consistent way to understand who is driving spend and which models are consuming budget. Previously, Vantage mapped costs from AI providers onto our VQL schema, which was built to normalize infrastructure costs and maps directly to the FinOps FOCUS schema. This approach did not account for the way AI pricing models diverge dramatically from traditional cloud and infrastructure providers. Each provider exposes different fields: the model name may appear as the service on one integration and as a tag on another; user identity may be an email, user ID, or API key depending on the vendor. For example, when procuring models through AWS, the Service field would remain AWS Bedrock, while the model name would appear in the subcategory alongside additional information such as region. By contrast, direct AI providers would use the model name itself as the Service field, making it difficult to have a clean report of model spend across multiple providers. Customers have tried to bridge these gaps with Virtual Tags, but those configs can become difficult to maintain as LLM vendors evolve and ship new models, token types, and billing fields.

Now, with Managed AI Tags, Vantage maps AI provider billing and usage fields into a consistent set of vntg:ai: tags. The same keys, such as vntg:ai:model, vntg:ai:model_provider, and vntg:ai:token_type, will appear whether spend comes directly from providers such as Cursor, Anthropic, OpenAI, or a cloud provider like AWS or Azure. As part of this, Vantage also extracts the dimensions customers actually want to filter on from messy provider-native labels. For example, GCP Marketplace Claude may bill a line as Claude Sonnet 4 — Output Tokens — Context Window Size from 0 to 200000 Tokens; with Managed AI Tags, customers can filter on vntg:ai:token_type is output and get the same dimension across providers. Customers using AI Gateway Enrichment also get request metadata, such as user email or endpoint, written into the same schema when billing data alone does not expose those fields. This lets customers build Budgets, Cost Reports, and Alerting on the same dimensions of cost across providers. Vantage maintains an ongoing, exhaustive map of all supported providers to normalized attributes to group costs across an AI vendor. This allows for evolving reports as you consume new SKUs that are consistent across AI, LLM, and inference providers.

Managed AI Tags will be available to customers with supported AI cost integrations. To get started, open a Cost Report and group or filter by any vntg:ai: tag key. Learn more in the Vantage documentation.

Frequently Asked Questions

1. What is being launched today?

Vantage is launching Managed AI Tags, a Vantage managed vntg:ai: tag schema applied automatically to AI cost so customers can analyze spend with consistent dimensions across providers.

2. Who is the customer?

Organizations tracking AI spend across one or more providers (for example Cursor, Anthropic, OpenAI, or AWS Bedrock) who need consistent allocation by model, user, token type, or related attributes.

3. How much does this cost?

There is no additional cost for Managed AI Tags. The feature is included with Vantage for accounts with supported AI integrations.

4. What tags are included in the Managed AI Tags schema?

Tag key	What it tells you	Example values
vntg:ai:model	The cleansed, normalized model name, allowing users to compare the same model across providers and release dates	gpt-4o, claude-sonnet-4
vntg:ai:raw_model	The exact model name from the provider	gpt-4o-2024-08-06, claude-sonnet-4-20250514
vntg:ai:model_provider	Who built the model—for example OpenAI or Anthropic—even if another platform hosts or bills for it, used to standardize models when procured through cloud providers	Anthropic, OpenAI, Google, xAI
vntg:ai:token_type	What kind of tokens this charge is for.	input, output, cache_read, cache_write
vntg:ai:provider_region	Where the request ran—region or geography.	us-east-1, global
vntg:ai:service_tier	The pricing or speed tier for this usage.	default, priority, batch, flex
vntg:ai:api_key_id	Which API key was used.	Provider key ID
vntg:ai:user_email	Email of the person who triggered the usage.	alice@company.com
vntg:ai:user_name	Display name of the person who triggered the usage.	Alice Smith
vntg:ai:user_id	User ID from the provider.	Provider user ID
vntg:ai:principal_id	Provider-specific caller identifier, that is not related to a specific username or email.	team-a/session

Not every provider supplies every field. Vantage only writes a tag when the source data provides a reliable value for that key.

5. Which providers are supported at launch?

Managed AI Tags are emitted automatically for supported AI cost integrations, including:

• Direct AI providers: Cursor, Anthropic (Admin API and Analytics), OpenAI, Anyscale, ElevenLabs, Baseten, Fireworks AI, and xAI
• Hyperscaler AI services: AWS Bedrock, Azure OpenAI, and Google Cloud AI

6. How does this differ from provider-specific tags like cursor: or anthropic:?

Provider-specific tags continue to exist and capture fields unique to that integration. Managed AI Tags add a cross-provider layer on top: the same vntg:ai:model or vntg:ai:user_email key works whether the spend came from Cursor or Anthropic.

Provider-specific native tags still appear alongside managed tags in cost data.

7. How does this differ from LLM Token Enrichment?

LLM Token Allocation joins external observability or gateway telemetry onto cost data to fill gaps that billing data alone does not provide (for example team or application metadata from Cloudflare AI Gateway or CloudWatch). Managed AI Tags normalize fields that are already present in supported provider cost and usage data into a shared schema. The two features are complementary: Managed AI Tags standardize what providers already send; Token Allocation enriches what providers leave out.

8. Can I change or override the schema?

Vantage maintains the canonical Managed AI Tag definitions so they stay consistent across accounts with the set prefix of vntg:ai:. If customers wish to change the naming conventions of tags, they can utilize the Collapsed Virtual Tag feature.

9. Will this break my existing Virtual Tags, Billing Rules, or reports?

No. Existing Virtual Tags, Billing Rules, and provider-specific tags are not modified. Managed AI Tags are added to cost rows automatically. You can use them in new reports and filters alongside your existing configuration.

10. How do I use Managed AI Tags in Cost Reports?

Open a Cost Report that includes AI provider spend, then filter or group by any available vntg:ai: tag. For example group by vntg:ai:model, or filter vntg:ai:user_email to a specific developer.

11. Can I backfill these tags?

No, normalized vntg:ai: tags will be available August 2026 and onwards.

12. Can I create Virtual Tags on top of Managed AI Tags?

Yes. Managed AI Tags are a foundation for cross-provider AI attribution. You can build Virtual Tags, Budgets, and Cost Alerts that reference vntg:ai:* keys the same way you use any other tag.