All insights

Model APIs

Gemini Code Assist CLI: Access, Features, and the Gemini API

Learn what “Gemini Code Assist CLI” refers to, which access routes may still work, how quotas differ from API billing, and how to make a Gemini API request.

Illustration for Gemini code assist cli

“Gemini Code Assist CLI” is not the product name used in Google’s current documentation: the terminal agent is Gemini CLI (gemini). It is distinct from Gemini Code Assist’s IDE agent mode and from a Gemini API integration in your application. Google documents that Code Assist editions provide CLI quotas shared with Code Assist agent mode; its Cloud page also describes Code Assist agent mode in VS Code as powered by Gemini CLI. (Google Cloud CLI documentation)

Which route fits your task?

Need Route Access condition
An agent in a terminal for project work Gemini CLI, if your account has an eligible route Google’s current CLI guidance describes organizational Code Assist access and certain legacy paid API or Vertex AI key routes. Do not assume a new API key alone grants CLI access; check the authentication guide.
Coding help inside an IDE Gemini Code Assist The individual/free and Google AI Pro/Ultra routes stopped serving requests through Gemini Code Assist IDE extensions on June 18, 2026. The transition announcement directs those users to Antigravity. (Google Developers Blog)
Model responses inside your own application Gemini API or an appropriate Google Cloud model API Your application sends model requests and handles credentials, billing, rate limits, and errors. An API request is not a repository-editing agent workflow.
An individual terminal-agent workflow Antigravity CLI Google’s transition announcement directs individual/free and Google AI Pro/Ultra users there.

Access caveat for organizations: Google’s Gemini Code Assist overview says new Standard and Enterprise subscriptions are no longer available for purchase. An organization with existing eligible access may still have a route, but a new reader should not assume they can simply buy one. (Gemini Code Assist overview)

What Gemini CLI does—and what access means

Google Cloud describes Gemini CLI as an open-source agent for a local terminal or Cloud Shell. Its documentation names bug fixes, feature creation, test coverage, research, and content generation as use cases, and describes built-in tools and local or remote MCP servers. These are documented capabilities, not guarantees of correct or safe changes. The available Cloud page is in Spanish and warns that AI-generated translations may contain errors. (Google Cloud CLI documentation)

For an eligible user, the CLI quick start gives the current installation command:

npm install -g @google/gemini-cli
 gemini

Use the authentication guide to check the sign-in or key method and project requirements for your account. Authentication instructions do not, by themselves, establish that an account is eligible for Gemini CLI access.

Google says Code Assist edition quotas for Gemini CLI are shared with Code Assist agent mode. Its quota documentation is the destination for current Standard and Enterprise conditions. I could not establish from the available evidence a reliable numeric limit or reset schedule applicable to every eligible edition and account, so check the current quota details for the specific account rather than relying on older preview figures. Google’s API rate limits are a separate system: they are project-specific, measured across dimensions such as requests per minute, input tokens per minute, and requests per day; API daily request quotas reset at midnight Pacific time. (Gemini API rate limits)

An August 2025 Google Developers Blog post described a particular Gemini CLI–VS Code integration with workspace, open-file, and selected-text context, plus reviewable, editable diffs. Its version requirement and /ide commands are historical instructions for that integration—not confirmation of current setup or a universal CLI feature. (Google Developers Blog)

Quotas, API billing, and privacy

Do not treat Code Assist quotas and direct Gemini API billing as interchangeable. The Cloud CLI page describes shared Code Assist quotas for Gemini CLI and agent mode, while Gemini API calls use the project’s API billing and rate-limit arrangements. For API billing, Google documents a minimum $5 USD (or equivalent) prepayment to move from the Free Tier to a Paid Tier. (Gemini API billing)

For privacy, the Cloud CLI documentation scopes Code Assist data-protection practices to CLI use by Standard and Enterprise users; review the applicable plan terms for those accounts. For paid Gemini API services, Google says prompts and responses are not used to improve its products, but that is not a zero-retention guarantee: the data-use and retention documentation describes limited logging for abuse monitoring and additional retention conditions for some features. Review the terms for your chosen route and features; do not infer that one route satisfies enterprise processing requirements without checking its applicable terms.

Minimal Gemini API example

A model API is the route to consider when your application needs model output. Google says Gemini 4 Argon (gemini-4-argon) is available through the Gemini API and AI Studio and is built for complex coding and agentic tasks. The current model guide documents an Interactions API request; this minimal REST example follows that documented pattern. Create a key using Google’s API-key guide, keep it out of source code, and supply it as an environment variable:

export GEMINI_API_KEY="your-api-key"
curl "https://generativelanguage.googleapis.com/v1beta/interactions" \
  -H "x-goog-api-key: $GEMINI_API_KEY" \
  -H "Content-Type: application/json" \
  -X POST \
  -d '{"model":"gemini-4-argon","input":"Summarize the purpose of this service in one sentence."}'

This is based on Google’s published request example; it is not a claim that the request was tested here. For next steps, use the model page for model details, the pricing page for current rates, and the rate-limits guide for the project-specific lookup. Google says you can view active limits in AI Studio; limits apply per project, not per API key. The exact limits vary by model and project. (Gemini API rate limits)

Recommendation

  • Already have eligible organizational Code Assist access? Check the current authentication and quota guidance for your account. CLI and agent-mode quotas are shared; the applicable numeric limits and reset conditions should be confirmed there.
  • Using an individual account or Google AI Pro/Ultra for an interactive terminal agent? Follow Google’s Antigravity CLI transition guidance; the Gemini CLI and Code Assist IDE consumer transition took effect June 18, 2026.
  • Building an application that calls a model? Use a model API directly and check that project’s model availability, billing, rate limits, and privacy terms.

Selection rule: CLI for an eligible terminal-agent workflow, IDE assistance when your account has a supported IDE route, and an API for model inference in your application.

Sources

An important update: Transitioning Gemini CLI to Antigravity CLI

Gemini Code Assist Standard and Enterprise overview

CLI de Gemini — Google Cloud documentation

Get started with Gemini CLI

Gemini CLI authentication setup

Quotas and limits — Gemini for Google Cloud

Rate limits — Gemini API

Billing — Gemini API

Zero data retention in the Gemini Developer API

What's new in Gemini 4 Argon

Using Gemini API keys

Gemini Developer API pricing

Gemini CLI + VS Code: Native diffing and context-aware workflows

Contact us