[Automated] Draft docs (agentgateway): llm: support service tier pricing

[github-actions]

agentgateway/agentgateway#3661 — llm: support service tier pricing

Docs versions: standalone/main, kubernetes/main

No code-path rule covers 2 touched file(s). The planner could route these only by identifier search, so the candidate pages may miss what they changed, including in the other mode:

• crates/llm/src/types/gemini.rs
• crates/llm/src/types/vertex_gemini.rs

What changes for users: Model catalogs can set service-tier-specific pricing with tiers[].serviceTier, using normalized values standard, flex, priority, and reserved. agctl catalog import now includes vertex-pricing by default, so generated catalogs can include Vertex AI Gemini service-tier rates.

What the docs now say: The standalone and Kubernetes model-cost docs now list vertex-pricing as a default catalog import source. The shared catalog-format snippet now explains service-tier tier rules and shows a serviceTier example for the development tree.

Not verified: No cluster or local runtime was used; the documented configuration and commands are unrun.

How agentgateway-3661 was drafted, and what was not verified

• Code PR: llm: support service tier pricing agentgateway#3661
• Code issue: none linked
• Product: agentgateway (upstream) → agentgateway/website
• Docs version: standalone/main, kubernetes/main; documented in the development tree, which is where unreleased changes go.
• Category: scoped-edit / pr

Plan, and what changed it

• Planned: Update the model-cost documentation for service-tier pricing and the new catalog import source.
• Learned: Generated schema and agctl reference files already reflect the diff but are generator-owned; the editable user-facing pages are the model-cost pages and shared catalog-format snippet.
• Did: Documented serviceTier pricing in the shared catalog-format snippet, added vertex-pricing to the standalone and Kubernetes cost guides, and wrote a proposed feature release note.

Why a documentation change is necessary

The diff adds serviceTier to model catalog tiers[], normalizes provider-reported service tiers to standard, flex, priority, or reserved, and uses the served tier when selecting effective rates. The existing model-cost pages described only context-length tiers, so a reader could not discover how to price Flex, Priority, or Reserved requests. The diff also adds vertex-pricing to the default import sources, but both cost guides still listed only models.dev and aws-bedrock-mantle.

What changed on disk

• assets/agw-docs/snippets/model-catalog-json-format.md: Explains service-tier-specific catalog tiers and shows a gated serviceTier example for the development tree.
• assets/agw-docs/standalone/cost-catalog.md: Adds vertex-pricing as a main-only default import source for standalone cost catalog docs.
• content/docs/kubernetes/main/documentation/llm/cost-controls/costs.md: Adds vertex-pricing to the Kubernetes main cost catalog source list.

How to verify this change

1. Generate a catalog with the default sources.

   agctl catalog import --pretty --out ./catalog.json

2. Inspect a Vertex AI Gemini model in catalog.json. The model entry includes tiers[] entries with serviceTier values such as flex or priority.

3. Load a catalog that contains tiers[].serviceTier: flex and send a request that the provider serves under the Flex tier. The request cost fields, such as llm.costRates, use the Flex tier rates instead of only the context-length rates.

What was not verified

The configuration was not applied to a cluster, so every command in this diff is unrun.

The agctl catalog import command was not run because the sandbox has no network access, and the vertex-pricing source fetches remote Google Cloud pricing data.

How the pages were chosen (triage report)

Draft a documentation change

agentgateway/agentgateway#3661 — llm: support service tier pricing

• Product: agentgateway (upstream) → agentgateway/website
• Docs version: standalone/main, kubernetes/main
• Summary (read from the body): * Record service tier for openai (existing), anthropic (existing), bedrock (new), vertex (new), gemini (new) * Normalize service tier we expose * In model catalog, allow tiers to include service tier * In model catalog, pull vertex tier prices
• Not tested: no cluster was available, so every command in this diff is unrun
• No rule covers 2 touched file(s). The candidate pages below come from the files that DID match a rule, so they may scope this narrower than the change is. Read these before deciding the change is only about the pages listed:
  • crates/llm/src/types/gemini.rs
  • crates/llm/src/types/vertex_gemini.rs
• Also has a file for rates, not drafted (2):
  • content/docs/kubernetes/main/documentation/llm/cost-controls/optimize.txt (lists the neighbours, add standard)
  • content/docs/standalone/main/documentation/llm/cost-controls/optimize.txt (lists the neighbours, add standard)

Release notes

This pull request also adds its release-note entries:

• content/docs/kubernetes/main/release-notes/release-notes.md: Service-tier pricing in the model catalog (feature)
• content/docs/standalone/main/release-notes/release-notes.md: Service-tier pricing in the model catalog (feature)

Check these release-note links and names. Each note is on its page, but a link may not reach the drafted content there, or a name may belong to the other mode:

• content/docs/standalone/main/release-notes/release-notes.md: llm: support service tier pricing agentgateway#3661's note names vertex-pricing, which content/docs/kubernetes/main mentions and content/docs/standalone/main does not. Check that this tree's readers can set it, or word the note for this mode.

Draft branch: pr-tracker-draft-agentgateway-3661

[cloudflare-workers-and-pages]

Deploying agentproxy with    Cloudflare Pages

Latest commit:	1f0a023
Status:	✅  Deploy successful!
Preview URL:	https://919d53bd.agentproxy.pages.dev
Branch Preview URL:	https://pr-tracker-draft-agentgatewa-7x9o.agentproxy.pages.dev

View logs