copilot clit research about gemini for google cloud cloudaicompanion.googleapis.com quotas and system limts understanding

research.md

Google Cloud Platform — Gemini for Google Cloud (cloudaicompanion.googleapis.com) Quota & System Limit Reference

Focus: Gemini Code Assist Standard + Gemini CLI

Report scope: Every quota and system-limit row visible in the Cloud Console for the cloudaicompanion.googleapis.com API (the Gemini for Google Cloud / Code Assist backend), interpreted in the context of a user running Gemini Code Assist Standard tier via Gemini CLI. Sources are official Google documentation and the open-source Gemini CLI repository unless otherwise noted.

---

1. Executive Summary

The cloudaicompanion.googleapis.com quota page in the Google Cloud Console lists a mix of currently enforced product quotas, legacy backend labels carried forward from the "Duet AI" era, and feature-specific limits for capabilities (BigQuery Data Insights, Looker, Code Repository Index) that are orthogonal to the terminal-based Gemini CLI workflow.

For a Gemini Code Assist Standard subscriber using Gemini CLI, only two quota rows are practically consequential:

1. API requests per minute per user (120 req/min) — the short-burst rate limiter that applies to all Gemini CLI and agent-mode interactions.
2. The daily pool for Gemini CLI + agent mode — officially documented as 1,500 requests/day for Standard; 2,000/day for Enterprise.¹² If the Cloud Console displays 2,000/day for a Standard project, treat the console value as the live enforced limit; the 1,500 figure is the published default.

One important nuance: a single user-visible prompt in Gemini CLI may internally generate multiple model requests (tool calls, context window refills, agent steps), each of which counts against both the per-minute and daily limits.¹

All other rows in the console — Anarres, Data Insights, Duet Complete/Generate/Task/Text, Code Repository Index, Generate Looker Query — either govern unrelated GCP features or are legacy internal labels that have no direct user-facing meaning for Gemini CLI today.

---

2. What Matters Most for Gemini CLI + Code Assist Standard

Priority	Quota / Limit	Default value	Why it matters
Critical	Gemini CLI + agent mode daily pool	1,500 req/day (Standard)1 · 2,000/day (Enterprise)1	Every LLM call made by gemini in the terminal (including agent sub-steps) counts against this pool
Critical	Per-minute rate limit	120 req/min (tabulated in localized docs1; per-minute limiting confirmed in English docs1)	Burst protection; hitting it yields HTTP 429 — the CLI retries automatically3
Irrelevant	Code Repository Index limits	1 resource; 600 req/min	Enterprise code-customization admin; not used by Standard or CLI
Irrelevant	Anarres / Data Insights / Looker quotas	100–120 req/min	BigQuery / Looker product features; not invoked by Gemini CLI
Background	Chat / Code / Duet* day limits	240–6,000 req/day	IDE-oriented or legacy labels; not the Gemini CLI daily bucket per public docs

Console vs. docs discrepancy: If your Cloud Console shows 2,000 requests/day while you hold a Standard subscription, the console value is the live, project-specific enforced figure and should be trusted for practical purposes. The published 1,500/day default for Standard may have been quietly adjusted, or your project may have an Enterprise entitlement. Always treat the console as authoritative for operations; use the docs as a baseline reference.⁴⁵

---

3. Quota / System-Limit Table

Each row from the user's Cloud Console view is covered in the subsections below.

---

3.1 Number of Code Repository Index resources

Field	Value
Console name	Number of Code Repository Index resources
Console classification	System limit
Default value	1
Official / public mapping	Partially mapped. "Code Repository Index" is the indexing backend for the code customization feature.67 No dedicated quota reference page exists for this specific row name; the closest official pages are the code-customization overview6 and setup guide7.
What it governs	The maximum number of Code Repository Index objects (repository indexes used to build per-org code context) a project may create.
Relevance to Gemini CLI + Standard	None. Code customization / Code Repository Index is an Enterprise-only feature.6 Standard-tier users cannot create repository indexes, so this system limit is never reached.
Quota vs. system limit	Correctly classified as a system limit — Google's documentation defines system limits as fixed, non-adjustable maximums, in contrast to quotas which are adjustable.89 This value cannot be increased via a quota-increase request.
Canonical source	Code customization overview6; quota terminology8

---

3.2 Anarres API requests per minute per user

Field	Value
Console name	Anarres API requests per minute per user
Console classification	Quota
Default value	100 req/min
Official / public mapping	No public canonical page maps the name "Anarres." "Anarres" appears to be an internal Google code name for a feature or subsystem surfaced in the cloudaicompanion.googleapis.com quota namespace. It is not defined in the public Gemini quotas page10, the Code Assist quotas page11, or any other official product documentation found at time of writing.
What it governs	Unknown with certainty. Given its placement alongside Data Insights and Looker rows, it most plausibly gates a BigQuery or analytics-adjacent conversational feature. Treat as an internal/legacy label.
Relevance to Gemini CLI + Standard	None. Even if the Anarres system is active, it is not invoked by Gemini CLI terminal interactions.
Quota vs. system limit	Classified as a quota (adjustable in principle), though the internal name makes a quota-increase request unlikely to succeed without Google support context.5
Canonical source	No dedicated public page. Closest official context: Gemini quotas page10; Cloud quota overview4.

---

3.3 Generate Looker Query API requests per minute per user

Field	Value
Console name	Generate Looker Query API requests per minute per user
Console classification	Quota
Default value	100 req/min
Official / public mapping	Maps to LookML generation / Gemini in Looker capabilities. The Gemini in Looker overview12, the Conversational Analytics API overview13, and the LookML with Gemini page14 describe AI-assisted LookML and query generation. The Conversational Analytics API quotas page15 provides the closest quota reference for Looker-related AI quotas.
What it governs	Rate-limits Looker Explore and LookML generation requests routed through the cloudaicompanion backend — for example, when Looker uses Gemini to auto-generate SQL or LookML.1416
Relevance to Gemini CLI + Standard	None. Gemini CLI does not interface with Looker Explore or LookML generation. This quota is only relevant if your organization uses Gemini-powered Looker features.
Quota vs. system limit	Quota — adjustable.4
Canonical source	Conversational Analytics API quotas15; Gemini in Looker overview12.

---

3.4 API requests per minute per user

Field	Value
Console name	API requests per minute per user
Console classification	Quota
Default value	120 req/min
Official / public mapping	This is the primary per-minute rate limit for Gemini CLI and agent-mode interactions. The Gemini CLI documentation2 confirms per-minute rate limiting in prose; a localized version of the same page tabulates 120 req/min explicitly. The Gemini quotas page10 and Code Assist quotas page11 both describe per-minute limits for Code Assist features.
What it governs	Short-burst throttle on API calls from any single user (identified by Google Account/OAuth token) against the cloudaicompanion endpoint. Gemini CLI authenticates via OAuth2 using the cloud-platform scope17 and all generateContent / streamGenerateContent calls to https://cloudcode-pa.googleapis.com/v1internal3 count against this limit. Agent-mode sub-steps each count as individual requests.
Relevance to Gemini CLI + Standard	High. This is the short-term throttle you will hit if a complex agentic task issues many rapid LLM calls. The CLI source code (server.ts) retries on HTTP 429 with a 1-second delay for up to 3 attempts3, so brief bursts may succeed transparently, but sustained high-frequency usage will be capped.
Quota vs. system limit	Quota — adjustable via a quota-increase request5, though increases for per-user/per-minute dimensions require Google support approval.
Canonical source	Gemini CLI docs2; Gemini quotas page10; Code Assist quotas page11.

---

3.5 Data Insights API requests per minute per user

Field	Value
Console name	Data Insights API requests per minute per user
Console classification	Quota
Default value	120 req/min
Official / public mapping	Maps to BigQuery Data Insights and the Generate Table Insights feature.1819 The BigQuery data canvas page20 and the Gemini in BigQuery overview21 provide broader context. No dedicated quota page exists for this specific row name; the Gemini quotas page10 is the closest authoritative source.
What it governs	Rate-limits per-user calls to the Data Insights AI feature within BigQuery — for example, auto-generated schema summaries and table insights surfaced in the BigQuery console.
Relevance to Gemini CLI + Standard	None. Gemini CLI does not invoke BigQuery Data Insights. This limit only affects users interacting with the BigQuery console or API's Gemini-powered insights features.
Quota vs. system limit	Quota — adjustable.4
Canonical source	BigQuery Data Insights docs1819; Gemini quotas page10.

---

3.6 Duet Text API requests per day per user

Field	Value
Console name	Duet Text API requests per day per user
Console classification	Quota
Default value	240 req/day
Official / public mapping	Legacy label. "Duet Text" is a backend quota category inherited from the Duet AI era (Google's predecessor brand to Gemini for Google Cloud). Current public documentation does not use this name. The Gemini quotas page10 and Code Assist quotas page11 have largely consolidated these into broader Gemini quota categories. The 240/day figure is low relative to other categories, suggesting this covers a narrow text-generation sub-feature.
What it governs	Likely governs text-generation interactions outside of chat/code completion — possibly inline explanations, commit-message generation, or similar short-text features in IDE plugins. The exact current mapping is uncertain.
Relevance to Gemini CLI + Standard	Low to unknown. Gemini CLI's primary interaction path uses streamGenerateContent against the general Gemini model endpoint3, not a "Duet Text" sub-API. If this bucket is still actively enforced, it could in theory cap certain in-IDE features, but it does not appear to be the binding daily limit for CLI sessions.
Quota vs. system limit	Quota — adjustable. The low 240/day default is notable and may indicate this is a seldom-used or deprecated path.
Canonical source	No dedicated public page for this exact label. Closest: Gemini quotas page10; Code Assist quotas page11. Mark as legacy/internal label.

---

3.7 Code Repository Index Management API requests per minute

Field	Value
Console name	Code Repository Index Management API requests per minute
Console classification	System limit
Default value	600 req/min
Official / public mapping	Maps to the code customization admin/management API — the operations used to create, update, delete, and poll the status of Code Repository Index resources.67 No dedicated quota reference page exists for this row name; the code-customization setup guide7 is the closest.
What it governs	Rate of management-plane API calls (create/delete/list/update index) to the Code Repository Index service. This is an admin-level limit, not a per-inference limit.
Relevance to Gemini CLI + Standard	None. Standard-tier users cannot use Code Repository Index at all.6 Even for Enterprise users, this governs provisioning operations performed by admins, not developer CLI sessions.
Quota vs. system limit	Correctly classified as a system limit — fixed and non-adjustable.8
Canonical source	Code customization overview6; code customization setup7; quota terminology8.

---

3.8 Duet Task API requests per day per user

Field	Value
Console name	Duet Task API requests per day per user
Console classification	Quota
Default value	960 req/day
Official / public mapping	Legacy label. "Duet Task" is another backend quota category inherited from the Duet AI era. No current official documentation defines a "Duet Task API" surface. Public docs have largely collapsed these into broader Gemini categories.1011
What it governs	Likely governed multi-step or task-oriented features within Duet AI (predecessor to Gemini Code Assist). Possibly related to code actions, refactoring tasks, or inline code transformation features in IDE plugins.
Relevance to Gemini CLI + Standard	Uncertain / Low. The 960/day ceiling is above the per-user interactive rate but below the daily CLI pool, so it would not normally be a bottleneck. It is unclear whether Gemini CLI invocations consume "Duet Task" quota at all. Treat as a legacy/internal label with no confirmed current mapping.
Quota vs. system limit	Quota — adjustable.
Canonical source	No dedicated public page. Closest: Gemini quotas page10; Code Assist quotas page11. Mark as legacy/internal label.

---

3.9 Chat API requests per day per user

Field	Value
Console name	Chat API requests per day per user
Console classification	User provided as system limit (see note)
Default value	2,000 req/day
Official / public mapping	Partially mapped. The Gemini quotas page10 and Code Assist quotas page11 both describe daily chat/conversation limits. The 2,000/day figure aligns with the documented Enterprise daily CLI + agent-mode quota1, but the public docs state Standard gets 1,500/day.1
What it governs	Daily request budget for conversational (chat) interactions via the Code Assist backend — the type of interaction used by Gemini CLI's REPL loop and chat-mode IDE features.
Relevance to Gemini CLI + Standard	High. This row most likely represents the same daily pool that governs Gemini CLI usage. If it shows 2,000 for a Standard account, see §2 above for interpretation guidance. Trust the console value as the live enforced limit; the published default for Standard is 1,500.1
Quota vs. system limit	Note on user-provided classification: The user reported this as a system limit in the console, but Google's public documentation describes these daily limits as quotas (adjustable, not fixed system limits).10118 This may be a UI labeling artifact. Quotas can be increased via a quota-adjustment request5; system limits cannot.
Canonical source	Gemini quotas page10; Gemini CLI docs2; Code Assist quotas page11.

---

3.10 Code API requests per day per user

Field	Value
Console name	Code API requests per day per user
Console classification	User provided as system limit (see note)
Default value	6,000 req/day
Official / public mapping	Partially mapped. A 6,000/day limit for code-related requests is referenced in the Code Assist quotas page11 and the Gemini quotas page10 under code completion / code generation categories.
What it governs	Daily budget for code-related API calls — most likely IDE inline code completion (ghost text) and/or code generation calls triggered by IDE plugins (VS Code, JetBrains, etc.), not the Gemini CLI REPL.
Relevance to Gemini CLI + Standard	Low for CLI. This is the IDE code-completion pool, not the Gemini CLI daily bucket. A heavy IDE user could exhaust this independently of CLI usage. Gemini CLI's day-to-day requests are accounted against the Chat/CLI pool (§3.9), not this code-completion pool — though the backend architecture may share resources.
Quota vs. system limit	Note: User-reported as system limit, but documented as a quota (adjustable) in official docs.1011 Same UI-labeling caveat as §3.9.
Canonical source	Code Assist quotas page11; Gemini quotas page10.

---

3.11 Duet Complete Code API requests per day per user

Field	Value
Console name	Duet Complete Code API requests per day per user
Console classification	User provided as system limit (see note)
Default value	6,000 req/day
Official / public mapping	Legacy label. "Duet Complete Code" is the backend quota label for inline code completion from the Duet AI era. Current docs refer to this feature as code completion within Gemini Code Assist IDE plugins.2211 The 6,000/day figure aligns with modern Code Assist quota documentation.11
What it governs	Daily cap on single-token / inline code-completion requests triggered by the IDE plugin (e.g., VS Code autocomplete ghost text). Each keystroke-triggered suggestion counts as one request.
Relevance to Gemini CLI + Standard	Minimal for pure CLI users. Gemini CLI does not use the "Complete Code" surface; it uses chat/generation flows. This limit would only matter to Standard users who also use the IDE plugin heavily.
Quota vs. system limit	Note: User-reported as system limit; documented as a quota in official sources.1011
Canonical source	Code Assist quotas page11; Gemini quotas page10. Mark as legacy label for the name "Duet Complete Code".

---

3.12 Duet Generate Code API requests per day per user

Field	Value
Console name	Duet Generate Code API requests per day per user
Console classification	User provided as system limit (see note)
Default value	6,000 req/day
Official / public mapping	Legacy label. "Duet Generate Code" is the backend quota label for explicit code generation (e.g., "generate a function," inline chat-based code blocks, /fix or /explain commands) from the Duet AI era. Current docs describe this as code generation within Gemini Code Assist IDE features.2211
What it governs	Daily cap on explicit, prompted code-generation requests in IDE plugins — larger than completion requests, these are discrete "write me code" interactions.
Relevance to Gemini CLI + Standard	Low for CLI. Gemini CLI routes through the streamGenerateContent endpoint using the general Gemini model, not through a dedicated "Generate Code" sub-API.3 However, in some backend configurations, CLI code-generation calls may be counted here as well; this is not confirmed in public documentation.
Quota vs. system limit	Note: User-reported as system limit; documented as a quota.1011
Canonical source	Code Assist quotas page11; Gemini quotas page10. Mark as legacy label for the name "Duet Generate Code".

---

4. Key Interpretation Notes

4.1 Cloud Console Quota Rows vs. Public Product Documentation

The Cloud Console's quota page for cloudaicompanion.googleapis.com exposes the raw backend quota identifiers used by Google's internal infrastructure. These names do not always map one-to-one to product-level descriptions in public documentation. For example:

• Public docs say "Gemini CLI and agent-mode requests" — the console shows "Chat API requests per day per user" and/or "Code API requests per day per user".
• Public docs discuss code completion limits — the console shows "Duet Complete Code API requests per day per user".

There is no official Google document that provides a canonical mapping from every console quota row name to its product-level meaning. The Gemini quotas page¹⁰ and the Code Assist quotas page¹¹ are the closest official references but do not cover every row.

4.2 Quota vs. System Limit Distinction

Google's Cloud documentation defines these clearly⁸⁹:

• Quota: An adjustable usage limit. Default values are set by Google but can be raised by submitting a quota-increase request through the console or API.⁵ Examples: "API requests per minute per user", "Chat API requests per day per user".
• System limit: A fixed, non-adjustable constraint enforced by the service architecture. Cannot be increased regardless of account type or support tier.⁸ Examples: "Number of Code Repository Index resources", "Code Repository Index Management API requests per minute".

Important caveat: Several rows the user observed as "system limits" in the console correspond to values that the public documentation describes as quotas (e.g., daily Chat/Code/Duet request caps). This may reflect a UI labeling inconsistency in the console, a project-specific enforcement policy, or that some quotas are displayed as "system limits" when they are set to non-adjustable project defaults. When in doubt, check the "Edit quota" button availability in the console — if the button is present, the value is a quota; if it is absent or grayed out, it is being enforced as a system limit for your project.⁴

4.3 Why Some Rows Still Say "Duet"

Google rebranded "Duet AI for Developers" to "Gemini Code Assist" in early 2024. Backend quota identifiers (which are part of the service's internal API surface) were not fully renamed at the time of the product rebrand. As a result, the Cloud Console continues to surface quota labels like "Duet Complete Code," "Duet Generate Code," "Duet Task," and "Duet Text" — these are the same underlying quota buckets now marketed under the Gemini Code Assist brand.¹⁰¹¹ Users should treat any "Duet *" row as the functional equivalent of its current Gemini Code Assist counterpart feature.

4.4 Why Some Rows Are Unrelated to Gemini CLI

The cloudaicompanion.googleapis.com service is the shared backend API for all Gemini for Google Cloud features — not just Code Assist / CLI. This means a single API service umbrella includes:

• Gemini in BigQuery (Data Insights, table insights, data canvas)^18192021
• Gemini in Looker (Conversational Analytics, LookML generation, Generate Looker Query)^1213151416
• Code customization (Code Repository Index — Enterprise only)⁶⁷
• Gemini Code Assist (IDE plugins + Gemini CLI)²²²

All of these quota rows therefore appear on the same console page. A developer using only Gemini CLI will never consume quotas for the BigQuery or Looker rows, but they will be visible in the console.

---

5. Confidence Assessment

Claim	Confidence	Basis
1,500 req/day Standard / 2,000 req/day Enterprise for CLI + agent mode	High	Stated explicitly in Gemini CLI docs12
120 req/min per-user rate limit	Medium–High	Tabulated in a localized version of the CLI docs; per-minute limiting confirmed in English prose12; row present in console at 120
Chat API row = CLI daily pool	Medium	Inferred from alignment of 2,000/day figure and feature description; no explicit console-to-docs mapping is published
Code / Duet Complete / Duet Generate rows = IDE code-completion, not CLI	Medium	Inferred from feature names and separate documentation of IDE vs. CLI quotas11; no explicit statement in docs
Anarres = internal/unrelated to CLI	Medium	No public documentation for "Anarres"; placement alongside analytics rows suggests it is not a CLI resource
Duet Text / Duet Task = legacy labels, low CLI relevance	Medium	Consistent with rebrand timeline and low/modest daily limits; no explicit mapping in public docs
Code Repository Index limits = Enterprise only, irrelevant to Standard CLI	High	Explicitly stated in code customization docs6
BigQuery / Looker rows = unrelated to CLI	High	Orthogonal product features with separate documentation181912202113151416
Some "system limit" rows are actually quotas per Google's docs	High	Google's quota documentation clearly distinguishes quotas from system limits89; the daily caps are documented as quotas
Single prompt → multiple model requests	High	Explicitly documented in Gemini CLI / Code Assist docs1; confirmed by CLI source code architecture3

---

6. Footnotes

---

Report generated based on official Google Cloud documentation and open-source Gemini CLI repository sources. All quota values reflect defaults at time of research; values shown in your Cloud Console are always authoritative for your specific project and subscription tier.

Footnotes

1. Gemini quotas page — defines daily and per-minute limits for Gemini Code Assist features including CLI + agent mode. Standard: 1,500 req/day; Enterprise: 2,000 req/day; per-minute limiting stated in prose; 120 req/min tabulated in localized variant. https://cloud.google.com/gemini/docs/quotas ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹²

2. Gemini CLI documentation — describes the terminal-based Gemini CLI, its authentication model, quota behavior, and usage guidance. https://cloud.google.com/gemini/docs/codeassist/gemini-cli ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

3. Gemini CLI source — server.ts — confirms the backend endpoint (https://cloudcode-pa.googleapis.com/v1internal), API version, and retry logic on HTTP 429 (3 retries, 1-second delay for generateContent; immediate retry for streaming). google-gemini/gemini-cli commit a809bc7: https://github.com/google-gemini/gemini-cli/blob/a809bc7c513a95d859082fa9ea1f265c7fba17ee/packages/core/src/code_assist/server.ts#L68-L69 ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

4. Cloud quotas overview — Google's authoritative explanation of the quota system, how limits are applied, and how to view them in the console. https://cloud.google.com/docs/quotas/overview ↩ ↩² ↩³ ↩⁴ ↩⁵

5. Quota management and increase requests — how to view, edit, and request increases for adjustable quotas. https://cloud.google.com/docs/quotas/view-manage; https://cloud.google.com/docs/quotas/help/request_increase ↩ ↩² ↩³ ↩⁴ ↩⁵

6. Code customization overview — documents the Code Repository Index feature, confirms Enterprise-only availability. https://cloud.google.com/gemini/docs/codeassist/code-customization-overview ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

7. Code customization setup guide — step-by-step instructions for creating and managing Code Repository Index resources (Enterprise admins only). https://cloud.google.com/gemini/docs/codeassist/code-customization ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

8. Cloud quota terminology — official definitions of "quota" (adjustable) vs. "system limit" (fixed/non-adjustable). https://cloud.google.com/docs/quotas/terminology ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

9. Reiteration of system-limit vs. quota distinction; see ⁸. ↩ ↩² ↩³

10. Gemini for Google Cloud quotas page — primary official reference for all cloudaicompanion.googleapis.com quota values. https://cloud.google.com/gemini/docs/quotas ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²²

11. Code Assist quotas page (developers.google.com) — supplementary quota reference for Gemini Code Assist; covers daily and per-minute limits across IDE and CLI features. https://developers.google.com/gemini-code-assist/resources/quotas ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²² ↩²³

12. Gemini in Looker overview — product overview for AI capabilities within Looker. https://cloud.google.com/gemini/docs/looker/overview ↩ ↩² ↩³ ↩⁴

13. Conversational Analytics API overview — documents the API that underpins Looker's Gemini-powered natural-language query features. https://cloud.google.com/gemini/docs/conversational-analytics-api/overview ↩ ↩² ↩³

14. LookML with Gemini — documentation for using Gemini to write and modify LookML. https://cloud.google.com/looker/docs/write-lookml-gemini ↩ ↩² ↩³ ↩⁴

15. Conversational Analytics API quotas — quota reference for Looker conversational AI features; the closest official source for Looker-related cloudaicompanion quota rows. https://cloud.google.com/gemini/docs/conversational-analytics-api/quotas ↩ ↩² ↩³ ↩⁴

16. Gemini quick starts in Looker — quick-start guides for Looker's Gemini-powered features. https://cloud.google.com/looker/docs/gemini-quick-starts ↩ ↩² ↩³

17. Gemini CLI source — oauth2.ts — confirms OAuth2 authentication using the cloud-platform scope; OAuth client ID is a public "installed application" client (not a secret per Google's installed-app OAuth guidelines). google-gemini/gemini-cli commit a809bc7: https://github.com/google-gemini/gemini-cli/blob/a809bc7c513a95d859082fa9ea1f265c7fba17ee/packages/core/src/code_assist/oauth2.ts#L55-L80 ↩

18. BigQuery Data Insights — feature documentation for AI-generated table/dataset insights in BigQuery. https://cloud.google.com/bigquery/docs/data-insights ↩ ↩² ↩³ ↩⁴

19. BigQuery generate table insights — details on the "Generate Insights" button and API in BigQuery. https://cloud.google.com/bigquery/docs/generate-table-insights ↩ ↩² ↩³ ↩⁴

20. BigQuery data canvas — interactive data-exploration feature in BigQuery powered by Gemini. https://cloud.google.com/bigquery/docs/data-canvas ↩ ↩² ↩³

21. Gemini in BigQuery overview — product overview for all Gemini AI features within BigQuery. https://cloud.google.com/gemini/docs/bigquery/overview ↩ ↩² ↩³

22. Gemini Code Assist overview — product description, tier comparison (Standard vs. Enterprise), and feature list. https://cloud.google.com/gemini/docs/codeassist/overview ↩ ↩² ↩³

GPT-5.4 · high(21%

● Context Usage

  ○ ○ ○ ○ ○ ○ ○ ◉ ◉ ◉   gpt-5.4 · 63k/304k tokens (21%)
  ◉ ◉ ◉ ◉ ◉ ◉ ◉ ◉ ◉ ◉
  ◉ · · · · · · · · ·   ○ System/Tools:  21.1k (7%)
  · · · · · · · · · ·   ◉ Messages:      41.7k (14%)
  · · · · · · · · · ·   · Free Space:   195.7k (64%)
  · · · · · · · · · ·   ◎ Buffer:        45.6k (15%)

Changes   +0 -0
Requests  1 Premium (28m 52s)
Tokens    ↑ 6.6m • ↓ 96.4k • 6.0m (cached) • 4.2k (reasoning)