docs: Add Gemini Enterprise (Vertex AI) BYOLLM documentation - #407
docs: Add Gemini Enterprise (Vertex AI) BYOLLM documentation#407hongyi-chen wants to merge 6 commits into
Conversation
- New page: enterprise/enterprise-features/byollm-gemini-enterprise - Update BYOLLM overview page with supported providers section - Refresh stale 'AWS Bedrock only / Google Vertex coming soon' claims across pricing FAQs, BYOK, custom inference endpoint, enterprise billing, admin panel, security overview, and platform FAQs - Add sidebar entry Co-Authored-By: Oz <oz-agent@warp.dev>
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
![warp-agent-staging[bot]](https://avatars.githubusercontent.com/in/2196476?s=60&v=4){kind=small}
There was a problem hiding this comment.
Overview
This PR establishes a dedicated Gemini Enterprise BYOLLM page, places it correctly beside the existing BYOLLM overview, and adds useful cross-links across Enterprise, inference, and billing content. The WIF, IAM, model-routing, and credential-lifecycle descriptions align with the shipped client and server implementation; publication still needs release-state, billing, cloud-agent-support, and verified-UI-surface corrections.
Concerns
Messaging and accuracy — Clarify whether the public launch is intentional while the current production feature flag remains disabled, reconcile AI-credit and platform-credit language, and remove or verify the model-details UI claim.
Information architecture — The new page and sidebar placement are appropriate, and the AWS/Gemini pages now cross-link bidirectionally. Update the provider comparison table so its cloud-agent column does not imply that Gemini Enterprise supports cloud agents.
Tone and style — The new content follows the major frontmatter, header, callout, link, Settings-path, and second-person rules. The changed-file style lint reported glossary/unrecognized-term warnings and pre-existing hardcoded product-name variables; none identified a material new formatting defect.
Verdict
Checks: CI ✅ · internal links ✅ (3,204 checked) · changed-file style lint
Found: 0 critical, 3 important, 1 question
Request changes before publishing
Review run
https://oz.staging.warp.dev/runs/019faaca-2a9f-706b-bce4-d0b579f56ef3
| Gemini Enterprise is one of the providers supported by [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/). For AWS-based routing, see the [AWS Bedrock BYOLLM setup](/enterprise/enterprise-features/bring-your-own-llm/). | ||
|
|
||
| :::note | ||
| Gemini Enterprise BYOLLM is only available on Warp's Enterprise plan and requires Warp `v0.2026.06.17` or later. [Contact sales](https://www.warp.dev/contact-sales) to learn more. |
There was a problem hiding this comment.
❓ [QUESTION] Is this page intended to launch publicly now? The current production configuration keeps gemini_enterprise disabled, and v0.2026.06.17 is the client-compatibility threshold rather than an availability guarantee. Please confirm the release state; otherwise label the feature as preview/unavailable or hold the public page.
There was a problem hiding this comment.
we're going to launch this on thurs july 30
There was a problem hiding this comment.
Confirmed: the requester (hongyi-chen) has confirmed a public launch for Thursday, July 30. The docs are being prepared ahead of the flag flip intentionally. No doc change is needed — no 'coming soon' or feature-flag caveats will be added to the Gemini Enterprise page.
- team-managed-keys-and-endpoints.mdx: split BYOLLM row into AWS Bedrock (cloud agents: Yes) and Gemini Enterprise (cloud agents: No, interactive sessions only); qualify prose to state Gemini Enterprise is interactive-only in this iteration - billing.mdx: remove incorrect '80% reduced rate' credit claim; replace with correct model — customer-routed BYOLLM inference consumes no AI credits, local runs still consume platform credits - byollm-gemini-enterprise.mdx: remove unverified 'Inference via Gemini Enterprise Agent Platform' UI label; keep verified badge mention only The caution box on byollm-gemini-enterprise.mdx (interactive-only scope) and the FAQ entry were already accurate. platform/faqs.mdx already had the Bedrock vs. Gemini distinction for cloud agents. 80% grep: only one occurrence in the entire docs tree (billing.mdx:46), now corrected. Co-Authored-By: Oz <oz-agent@warp.dev>
There was a problem hiding this comment.
Final re-review — accepted
The rework at b6f32547def468d63663964c242c6e2e67e8cc95 resolves the four prior review items. I found no must-fix issues or follow-up polish.
Verified:
- Credits - No
80%claim remains insrc/content/docs/. The billing, BYOLLM overview, and Gemini Enterprise pages consistently state that customer-routed inference consumes no AI credits, provider inference is billed to the customer cloud account, and local runs still consume platform credits. - Cloud agents - The comparison table now distinguishes AWS Bedrock BYOLLM (Yes) from Gemini Enterprise BYOLLM (No, interactive sessions only). Related copy consistently scopes Gemini Enterprise to interactive requests and AWS Bedrock to current cloud-agent BYOLLM support.
- Model picker - The unverified model-details label was removed. The remaining references describe only the Gemini Enterprise badge without asserting an unverified UI string.
- Regression checks - The MDX table renders successfully in CI,
git diff --checkpasses, and no broken internal links were found.
Validation:
gh pr checks: passed (8/8).- Internal link checker: passed (3,207 internal links, 0 broken).
style_lint --changed: completed; reported 111 existing/non-blocking warnings, with no new material issue in this rework.- Local
npm run build: reached static-route generation, then hit the known/_llms-txt/support.txtstack-overflow failure under unsupported Nodev25.0.0; the PR's CI build/link/audit check passed.
Review run: https://oz.staging.warp.dev/runs/019faaca-2a9f-706b-bce4-d0b579f56ef3
|
I'm starting a first review of this pull request. You can view the conversation on Warp. I completed the review and no human review was requested for this pull request. Comment Powered by Oz |
![oz-for-oss[bot]](https://avatars.githubusercontent.com/in/3450139?s=60&v=4){kind=small}
There was a problem hiding this comment.
Overview
This PR adds a new Gemini Enterprise (Vertex AI) BYOLLM page and updates related Enterprise, billing, inference, and sidebar references to treat Gemini Enterprise as a supported BYOLLM provider. The structure and link placement are generally consistent with the docs, and no approved spec context was provided for a spec-drift pass.
Concerns
- The billing copy introduces
AI credits, which conflicts with the docs terminology standard to usecreditsfor the AI usage bucket. - The credential-security section says nothing sensitive leaves the customer's boundary, but the documented flow still sends request content and short-lived access tokens through Warp's backend; the claim needs to be narrowed to long-lived Google credentials.
Security
- Narrow the boundary claim so customers do not infer that prompts, responses, or access tokens never pass through Warp during Gemini Enterprise routing.
Verdict
Found: 0 critical, 3 important, 0 suggestions
Request changes
Comment /oz-review on this pull request to retrigger a review (up to 3 times on the same pull request).
Powered by Oz
| ## BYOLLM billing | ||
|
|
||
| When using [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/), Warp routes requests through your cloud infrastructure (AWS Bedrock today, with Azure Foundry and Google Vertex coming soon). BYOLLM requests **consume credits at a reduced rate** which is approximately 80% lower than standard usage. Inference costs are also billed directly to your cloud account. | ||
| When using [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/), Warp routes requests through your cloud infrastructure (AWS Bedrock and [Gemini Enterprise on Vertex AI](/enterprise/enterprise-features/byollm-gemini-enterprise/) today, with Azure Foundry coming soon). Customer-routed BYOLLM inference **consumes no AI credits** — Warp does not charge AI credits for requests that execute in your cloud. Inference costs are billed directly to your cloud account. Local agent runs that use BYOLLM still consume [platform credits](/support-and-community/plans-and-billing/platform-credits/) for Warp's platform infrastructure. |
There was a problem hiding this comment.
AI credits; use credits for the AI usage bucket so this page stays consistent with the rest of billing docs.
| When using [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/), Warp routes requests through your cloud infrastructure (AWS Bedrock and [Gemini Enterprise on Vertex AI](/enterprise/enterprise-features/byollm-gemini-enterprise/) today, with Azure Foundry coming soon). Customer-routed BYOLLM inference **consumes no AI credits** — Warp does not charge AI credits for requests that execute in your cloud. Inference costs are billed directly to your cloud account. Local agent runs that use BYOLLM still consume [platform credits](/support-and-community/plans-and-billing/platform-credits/) for Warp's platform infrastructure. | |
| When using [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/), Warp routes requests through your cloud infrastructure (AWS Bedrock and [Gemini Enterprise on Vertex AI](/enterprise/enterprise-features/byollm-gemini-enterprise/) today, with Azure Foundry coming soon). Customer-routed BYOLLM inference **consumes no credits** — Warp does not charge credits for requests that execute in your cloud. Inference costs are billed directly to your cloud account. Local agent runs that use BYOLLM still consume [platform credits](/support-and-community/plans-and-billing/platform-credits/) for Warp's platform infrastructure. |
There was a problem hiding this comment.
Addressed — updated to "consumes no credits" / "does not charge credits" on this line.
|
|
||
| When a request routes through Gemini Enterprise: | ||
|
|
||
| * **Warp doesn't consume AI credits** for that request. Inference is billed by Google Cloud to your project. |
There was a problem hiding this comment.
AI credits; use credits here to match the billing vocabulary.
| * **Warp doesn't consume AI credits** for that request. Inference is billed by Google Cloud to your project. | |
| * **No credits for inference** - Warp doesn't consume credits for that request. Inference is billed by Google Cloud to your project. |
| ### Credential security | ||
|
|
||
| * **No long-lived credentials** - The only long-lived credential involved is the member's Warp session. Access tokens are short-lived, held in memory, and never persisted. | ||
| * **Nothing sensitive leaves your boundary** - Warp stores only non-secret routing configuration (project ID, location, WIF audience, and optional service account email). Service account keys, refresh tokens, and credential files are never uploaded to Warp. |
There was a problem hiding this comment.
| * **Nothing sensitive leaves your boundary** - Warp stores only non-secret routing configuration (project ID, location, WIF audience, and optional service account email). Service account keys, refresh tokens, and credential files are never uploaded to Warp. | |
| * **No long-lived Google credentials leave your boundary** - Warp stores only non-secret routing configuration (project ID, location, WIF audience, and optional service account email). Service account keys, refresh tokens, and credential files are never uploaded to Warp. |
JaidenRatti
left a comment
JaidenRatti
left a comment
There was a problem hiding this comment.
Looks great, thank you for starting the first copy!! There are quite a few changes I'm going to make to the bulk of this. But left some comments on the non-meat part of this.
| | **Bring Your Own API Key** (BYOK) | Use your own API key for OpenAI, Anthropic, or Google models. Keys are stored locally on your device. | Free and all eligible paid plans | | ||
| | **[Custom inference endpoint](/agent-platform/inference/custom-inference-endpoint/)** | Connect Warp to an OpenAI-compatible endpoint such as OpenRouter, LiteLLM, z.ai, or an internal gateway. | Free and all eligible paid plans | | ||
| | **[Bring Your Own LLM](/enterprise/enterprise-features/bring-your-own-llm/)** (BYOLLM) | Enterprise-managed inference through your cloud provider (AWS Bedrock today; Azure Foundry and Google Vertex coming soon), with Warp handling routing, orchestration, governance, and observability. | Enterprise only | | ||
| | **[Bring Your Own LLM](/enterprise/enterprise-features/bring-your-own-llm/)** (BYOLLM) | Enterprise-managed inference through your cloud provider (AWS Bedrock and Gemini Enterprise on Vertex AI today; Azure Foundry coming soon), with Warp handling routing, orchestration, governance, and observability. | Enterprise only | |
There was a problem hiding this comment.
nit: this can be "and Gemini Enterprise Agent Platform". i dont think "on vertex" makes sense (though i think its fine if we add vertex in brackets like in custom-inference-endpoint.mdx since ppl prob still think of it as vertex).
There was a problem hiding this comment.
Done — updated to "Gemini Enterprise Agent Platform (Vertex AI)" here, and applied the same fix to the identical phrasing in custom-inference-endpoint.mdx and pricing-faqs.mdx for consistency.
|
|
||
| :::caution | ||
| BYOLLM currently supports **AWS Bedrock** only. Coming soon: Azure Foundry and Google Vertex support. | ||
| Azure Foundry support is coming soon. The rest of this page covers the AWS Bedrock configuration. |
There was a problem hiding this comment.
i think its a little confusing that the bring your own llm page explains the aws bedrock config - but its a separate page for just BYOLLM: Gemini Enterprise? Maybe they can both be their own pages?
There was a problem hiding this comment.
Agreed, and done — split AWS Bedrock out into its own page: src/content/docs/enterprise/enterprise-features/byollm-aws-bedrock.mdx, mirroring the Gemini Enterprise page's structure (key features, how it works, enabling BYOLLM + cloud agents, routing/fallback, billing, security, troubleshooting, FAQ). bring-your-own-llm.mdx is now a short BYOLLM overview/hub page: what BYOLLM is, a provider comparison table, and a general FAQ, linking out to both dedicated setup guides. Updated the sidebar and the AWS-specific cross-references in team-managed-keys-and-endpoints.mdx and byollm-gemini-enterprise.mdx accordingly.
| ## BYOLLM billing | ||
|
|
||
| When using [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/), Warp routes requests through your cloud infrastructure (AWS Bedrock today, with Azure Foundry and Google Vertex coming soon). BYOLLM requests **consume credits at a reduced rate** which is approximately 80% lower than standard usage. Inference costs are also billed directly to your cloud account. | ||
| When using [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/), Warp routes requests through your cloud infrastructure (AWS Bedrock and [Gemini Enterprise on Vertex AI](/enterprise/enterprise-features/byollm-gemini-enterprise/) today, with Azure Foundry coming soon). Customer-routed BYOLLM inference **consumes no AI credits** — Warp does not charge AI credits for requests that execute in your cloud. Inference costs are billed directly to your cloud account. Local agent runs that use BYOLLM still consume [platform credits](/support-and-community/plans-and-billing/platform-credits/) for Warp's platform infrastructure. |
There was a problem hiding this comment.
not related to gemini enterprise stuff specifically but wanted to callout here that it was intentional to remove this 80% callout to this new text instead?
There was a problem hiding this comment.
Yes, intentional — the flat 80% figure wasn't accurate (flagged in the first review round on this PR), so it was replaced with the correct model: customer-routed BYOLLM inference consumes no credits, and local agent runs still consume platform credits. While I was in there I also fixed "AI credits" → "credits" on this line (and added a link for AWS Bedrock) to match the terminology glossary — that lines up with the oz-for-oss suggestion on this same line too.
| Gemini Enterprise is one of the providers supported by [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/). For AWS-based routing, see the [AWS Bedrock BYOLLM setup](/enterprise/enterprise-features/bring-your-own-llm/). | ||
|
|
||
| :::note | ||
| Gemini Enterprise BYOLLM is only available on Warp's Enterprise plan and requires Warp `v0.2026.06.17` or later. [Contact sales](https://www.warp.dev/contact-sales) to learn more. |
There was a problem hiding this comment.
i think it's fine to remove the "requires Warp v0.2026.06.17. not really important here.
There was a problem hiding this comment.
Done — removed the v0.2026.06.17 requirement from the note.
|
|
||
| ## Key features | ||
|
|
||
| * **No local Google credentials required** - Members don't run `gcloud auth`, manage Application Default Credentials, or paste API keys. Warp mints a short-lived Google Cloud access token through Workload Identity Federation (WIF), rooted in each member's signed-in Warp session. |
There was a problem hiding this comment.
I'd prefer if we changed this to be something like
Session-based federated authentication — Warp uses the member’s signed-in Warp session to issue a short-lived OIDC identity token, then exchanges it through Google Workload Identity Federation for temporary Google Cloud credentials.
rather than focusing on what members dont run.
There was a problem hiding this comment.
Done — used your suggested wording verbatim: "Session-based federated authentication — Warp uses the member's signed-in Warp session to issue a short-lived OIDC identity token, then exchanges it through Google Workload Identity Federation for temporary Google Cloud credentials."
| * **Admin-controlled routing and models** - Admins configure the GCP project, Vertex location, and WIF provider once in the Admin Panel, then choose exactly which models are enabled and which Vertex model references they resolve to. | ||
| * **Gemini and Claude models** - Route native Gemini models and Claude partner models available on Vertex AI through your project. | ||
| * **Consolidated billing and quota attribution** - Inference runs against your project's Vertex AI quota and is billed to your Google Cloud account. Requests carry your project for quota attribution (`X-Goog-User-Project`). | ||
| * **No stored credentials** - Warp never stores long-lived Google credentials. Each request carries only a short-lived access token that Warp uses in memory and discards when the request ends. |
There was a problem hiding this comment.
maybe the text instead can be "Warp never stores service account keys, refresh tokens, or credential files. The client keeps only a short-lived Google Cloud access token in memory and automatically replaces it as it approaches expiration."
And the title instead can be "No long-lived credentials"
There was a problem hiding this comment.
Done — retitled the bullet to "No long-lived credentials" and used your suggested copy.
| * **Native Gemini models** - Current Gemini Flash, Flash Lite, and Pro families (for example, Gemini 3.1 Pro and Gemini 3.6 Flash). | ||
| * **Claude partner models on Vertex AI** - Current Claude Sonnet, Opus, Haiku, and Fable families offered as Vertex AI partner models. | ||
|
|
||
| In the Warp app's model picker, eligible models show a Gemini Enterprise badge when your credentials are active. Models absent from the registry (including extended-thinking and fast-mode variants, and Fireworks-hosted models like Kimi, MiniMax, Qwen, DeepSeek, and GLM) aren't available through Gemini Enterprise. Some models are disabled by default for Enterprise teams and require explicit admin enablement. |
There was a problem hiding this comment.
i think we can remove this.
There was a problem hiding this comment.
Done — removed.
|
|
||
| Here's the high-level flow: | ||
|
|
||
| 1. **Admin configures routing** - Your team admin enables the Gemini Enterprise host on the **Models** page of the [Admin Panel](/enterprise/team-management/admin-panel/) and sets the GCP project, Vertex location, and WIF provider. |
There was a problem hiding this comment.
"and WIF provider audience string" might be better here.
There was a problem hiding this comment.
Done — updated to end with "...and WIF provider audience string."
|
|
||
| * **Warp Enterprise plan with admin access** - You configure routing on the **Models** page of the [Admin Panel](/enterprise/team-management/admin-panel/). | ||
| * **A Google Cloud project with Vertex AI enabled** - Enable the Vertex AI API (`aiplatform.googleapis.com`) in the project that should own inference, quota, and billing. | ||
| * **Model access in Vertex AI** - Gemini models are available by default; Claude partner models must be enabled for your project in the Vertex AI Model Garden. |
There was a problem hiding this comment.
Not sure if this is actually the case, that gemini models are available by default for enterprise. Maybe instead just some generic copy like - ensure models you want to use in warp are enabled in your vertex.
|
|
||
| Setup has two halves: a one-time Google Cloud configuration (WIF trust and IAM), then routing configuration in Warp's Admin Panel. | ||
|
|
||
| ### Prerequisites |
There was a problem hiding this comment.
I'm going to manually update the rest of this file with a commit from this line onwards in this PR. Figured its easier than going back and forth.
There was a problem hiding this comment.
Understood — left everything from ### Prerequisites onward untouched for your rewrite, including the model-availability wording from your comment above. The only edits I made further down the file are two mechanical link-path updates (the #enabling-byollm-for-cloud-agents references at what are now lines ~20 and ~265), since that section now lives on the new byollm-aws-bedrock.mdx page after the AWS Bedrock split — no wording or content changes in your section.
…view feedback Splits BYOLLM: AWS Bedrock out of the Bring Your Own LLM overview page into its own dedicated page, mirroring the BYOLLM: Gemini Enterprise page structure, per review feedback that it was confusing for the overview page to fully document AWS Bedrock while Gemini Enterprise had its own page. - New page: enterprise/enterprise-features/byollm-aws-bedrock.mdx, with key features, how it works, enabling BYOLLM (interactive + cloud agents), routing/fallback, billing, security, troubleshooting, and FAQ sections moved out of bring-your-own-llm.mdx. - bring-your-own-llm.mdx is now a BYOLLM overview/hub page: what BYOLLM is, a provider comparison table, and general FAQ, linking out to the AWS Bedrock and Gemini Enterprise setup guides. - sidebar.ts: added the new page between "Bring your own LLM" and "BYOLLM: Gemini Enterprise". - team-managed-keys-and-endpoints.mdx: repointed AWS Bedrock BYOLLM links to the new page. - byollm-gemini-enterprise.mdx: applied Jaiden's requested wording changes (drop the specific version requirement, reword the credential bullets, tighten the WIF provider audience wording, drop the model-registry caveat paragraph) and fixed the AWS Bedrock cross-references that moved. Everything from "### Prerequisites" onward is intentionally left untouched, since Jaiden is rewriting that section directly. - Applied Jaiden's "Gemini Enterprise Agent Platform (Vertex AI)" wording nit consistently in bring-your-own-api-key.mdx, custom-inference-endpoint.mdx, and pricing-faqs.mdx. - billing.mdx: fixed "AI credits" -> "credits" terminology per the docs style guide, in the sentence Jaiden flagged. Validated with `npm run build`, the internal link checker (0 broken links), and `style_lint.py --all` (0 errors; remaining warnings are pre-existing proper-noun false positives, consistent with prior validation on this PR). Co-Authored-By: Oz <oz-agent@warp.dev>
| ### Does Warp support other model routers or "Bring Your Own LLM"? | ||
|
|
||
| On the Enterprise plan, Warp's [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/) lets you route inference through major cloud-provider Model-as-a-Service offerings. BYOLLM currently supports AWS Bedrock, with Azure Foundry and Google Vertex coming soon. Warp still manages model support, routing, and orchestration, but inference runs in your cloud environment so you can maintain data locality, security controls, and existing cloud spend commitments. | ||
| On the Enterprise plan, Warp's [Bring Your Own LLM (BYOLLM)](/enterprise/enterprise-features/bring-your-own-llm/) lets you route inference through major cloud-provider Model-as-a-Service offerings. BYOLLM supports AWS Bedrock and [Gemini Enterprise (Vertex AI)](/enterprise/enterprise-features/byollm-gemini-enterprise/), with Azure Foundry coming soon. Warp still manages model support, routing, and orchestration, but inference runs in your cloud environment so you can maintain data locality, security controls, and existing cloud spend commitments. |
There was a problem hiding this comment.
maybe link aws bedrock page here as well (if we are linking gemini page)
Rework the setup section so the console flow is the primary path and the gcloud CLI commands are a clearly scoped alternative. - Number the setup as steps 1-4 to match the AWS Bedrock page, and fix the broken step cross-references (a leftover "Step X" placeholder and a reference to an unnumbered "Step 2"). - Add a "Values you'll need" section defining PROJECT_ID, PROJECT_NUMBER, TEAM_UID, POOL_ID, and PROVIDER_ID once, with examples. - Split step 2 into Option A (direct federated access, recommended) and Option B (service account impersonation) and make the Admin Panel "Service account email" field conditional on that choice. - Clarify the WIF audience: canonical aud form in step 1, the // form for the Admin Panel field, plus a caution about the prefix and formatting. - Move the gcloud CLI commands into a scoped alternative section, unify placeholders on PROVIDER_ID, and add an export block so they run as-is. - Fix rendering bugs: a numbered list broken by intervening content and a table missing surrounding blank lines. - Point stale AWS Bedrock links at byollm-aws-bedrock and refresh troubleshooting to cover both access options. Co-Authored-By: Oz <oz-agent@warp.dev>
3 participants
Summary
Documents the newly shipped Gemini Enterprise (Vertex AI) BYOLLM support (REV-1599) and updates every page that claimed BYOLLM was "AWS Bedrock only, with Google Vertex coming soon." Content is sourced from
warp-server/specs/REV-1599/PRODUCT.md+TECH.mdand the client specwarp/specs/REV-1599/TECH.md.Changes
New page:
src/content/docs/enterprise/enterprise-features/byollm-gemini-enterprise.mdxhttps://app.warp.dev,sub/teamsclaim mapping, examplegcloudcommands, required IAM roles (roles/aiplatform.user,roles/serviceusage.serviceUsageConsumer), direct federated access vs. service account impersonation, and Admin Panel Models page configuration (project, location, WIF audience, optional SA, per-model refs, Enforce vs. Respect user setting).src/content/docs/enterprise/enterprise-features/bring-your-own-llm.mdxsrc/sidebar.tsStale claim refresh (Bedrock-only / "Google Vertex coming soon")
support-and-community/plans-and-billing/pricing-faqs.mdx(2 spots)agent-platform/inference/bring-your-own-api-key.mdx(comparison table)agent-platform/inference/custom-inference-endpoint.mdx(comparison table + centrally managed section)enterprise/support-and-resources/billing.mdx(BYOLLM billing)enterprise/team-management/admin-panel.mdx(Models settings)enterprise/index.mdx(key capabilities)enterprise/enterprise-features/team-managed-keys-and-endpoints.mdx(comparison + related resources)enterprise/security-and-compliance/security-overview.mdx(Models control + BYOLLM data locality)enterprise/getting-started/getting-started-enterprise.mdx(Admin Panel sections)platform/faqs.mdx(self-hosting LLM routing FAQ, now distinguishing Bedrock cloud-agent support from Gemini Enterprise's interactive-only scope)Validation
npm run buildpasses (351 pages, new page included)style_lint --changed: remaining flags on touched files are pre-existing or proper-noun false positives (e.g., "Gemini Enterprise", "Vertex AI", "ZDR" in headings, matching the existing Bedrock page's patterns)Notes for reviewers
gcloudexamples mirror the WIF contract in the specs (issuer,subshapescoped_principal:<team-uid>/<actor-type>:<principal-uid>,teamsclaim →google.groups); worth a skim from someone close to REV-1599 (cc Jaiden) to confirm the recommended IAM role set.Rework changes
Addressed code-review findings from review #4802286543:
Release state (❓ question — no doc change) — Confirmed with requester: public launch is Thursday July 30, docs intentionally prepared ahead of the flag flip. No "coming soon" or feature-flag caveats added. Thread replied and left open for reviewer acknowledgment.
Cloud agents — Gemini Enterprise interactive-only (⚠️ implemented, resolved) —
team-managed-keys-and-endpoints.mdx: split the single BYOLLM comparison-table row into two rows — AWS Bedrock (cloud agents: Yes) and Gemini Enterprise (cloud agents: No, interactive sessions only) — and rewrote the following prose to qualify cloud-agent support by provider.byollm-gemini-enterprise.mdxalready carried a correct caution box;platform/faqs.mdxalready distinguished the two providers. All other changed files were audited and found accurate.Credits — 80% claim removed (⚠️ implemented, resolved) —
billing.mdx:46: replaced the incorrect "reduced rate ≈ 80% lower" sentence with the correct model: customer-routed BYOLLM inference consumes no AI credits; inference billed to your cloud account; local runs still consume platform credits. Repo-wide grep for "80%": confirmedbilling.mdx:46was the only instance, now corrected.byollm-gemini-enterprise.mdxandbring-your-own-llm.mdxalready stated the correct model; no changes needed there.Model picker label — unverified string removed (⚠️ implemented, resolved) —
byollm-gemini-enterprise.mdx:179: removed the unverified "Inference via Gemini Enterprise Agent Platform" label string. The verified Gemini Enterprise badge mention is retained.Rework changes (round 2)
Addressed feedback from Jaiden's review on this PR:
bring-your-own-llm.mdxinto a BYOLLM overview/hub page plus a new dedicatedbyollm-aws-bedrock.mdxpage (mirroringbyollm-gemini-enterprise.mdx's structure: key features, how it works, enabling BYOLLM + cloud agents, routing/fallback, billing, security, troubleshooting, FAQ), since it was confusing for the overview page to fully document AWS Bedrock while Gemini Enterprise had its own page. Updatedsidebar.tsand the AWS-specific cross-references inteam-managed-keys-and-endpoints.mdxandbyollm-gemini-enterprise.mdx.bring-your-own-api-key.mdx, and applied the same fix to the identical phrase incustom-inference-endpoint.mdxandpricing-faqs.mdxfor consistency.byollm-gemini-enterprise.mdxwording (✅ implemented) — Dropped the specificv0.2026.06.17version requirement, reworded the "No local Google credentials required" and "No stored credentials" bullets to Jaiden's suggested copy ("Session-based federated authentication" / "No long-lived credentials"), tightened the WIF provider audience wording, and removed the model-registry caveat paragraph. Everything from### Prerequisitesonward is intentionally untouched — Jaiden is rewriting that section directly — aside from two mechanical link-path fixes for the#enabling-byollm-for-cloud-agentsanchor, which now lives on the new AWS Bedrock page after the split.billing.mdx80% callout (✅ confirmed intentional) — Confirmed the round 1 removal was intentional; also fixed "AI credits" → "credits" on that line per the terminology glossary.Validated with
npm run build, the internal link checker (0 broken links), andstyle_lint.py --all(0 errors; remaining warnings are pre-existing proper-noun false positives).Conversation: https://staging.warp.dev/conversation/08db3037-7ca3-4332-9318-c3e2a30766f5
Run: https://oz.staging.warp.dev/runs/019fb44b-f12a-7325-877a-4dfb7430d7a3
Originating thread: https://warpdev.slack.com/archives/C0BDQDW8V5E/p1785276411828319
Conversation: https://staging.warp.dev/conversation/50ee6770-15d6-457e-9f45-6f85a8094710
Run: https://oz.staging.warp.dev/runs/019faaa2-a7cd-762c-889b-b9b653c45541
This PR was generated with Oz.
Co-Authored-By: Oz oz-agent@warp.dev