
Choosing a Model in Antigravity
Choosing a model in Antigravity means deciding which reasoning model (Gemini, Claude or GPT-OSS) and which thinking level (low, medium, high, sometimes xhigh or max) will run the agent for each conversation, based on the Google AI package you're using and your remaining quota. This post doesn't rank which model is "best": Google doesn't publish such rankings in their documentation, and OPCAIOS hasn't measured quality or speed between models on the same task. Model names, thinking levels, commands and flags come from documentation at antigravity.google and from agy --help, agy models of CLI version 1.4.12, read on 03/11/2026. The model list changes quickly, so treat this post as a way to read and to adapt. General introduction to the application, IDE and CLI is in What is Antigravity.

Three knobs when choosing a model: model, thinking level and quota. Diagram: OPCAIOS.
What models does Antigravity have according to current documentation?
Antigravity has three main families of reasoning models according to the /docs/models page (read on 03/11/2026): Gemini from Google DeepMind, Claude from Anthropic and GPT-OSS 120B, with different usage rights per package. This page lists Gemini 4 Argon, Gemini 3.8 Flash, Gemini 3.7 Flash, Gemini 3.6 Flash, Gemini 3.1 Pro, Claude Sonnet 5.5 and Claude Opus 5.5 (Thinking edition), Claude Sonnet 4.6 and Claude Opus 4.6 (Thinking edition), plus GPT-OSS-120b. The table below summarizes usage rights by package, restated from the documentation:
| Model | Free and Google AI Plus | Google AI Pro | Google AI Ultra | Enterprise |
|---|---|---|---|---|
| Gemini 4 Argon | No | Yes (Pro paid package) | Yes | Yes |
| Gemini 3.8 Flash | Yes | Yes | Yes | Yes |
| Gemini 3.7 Flash, 3.6 Flash, 3.1 Pro | No | No | No | Yes |
| Claude Sonnet 5.5, Opus 5.5 | No | Yes (trial edition not included) | Yes | Yes |
| Claude Sonnet 4.6, Opus 4.6 | Yes | Yes | No | No |
| GPT-OSS-120b | Yes | Yes | Yes | No |
Most important note: Claude Sonnet 4.6, Claude Opus 4.6 and GPT-OSS-120b will be removed on 18/11/2026. Antigravity also uses Nano Banana 2 for image generation tools, which part cannot be chosen.
What does the agy models command show on a specific account?
The agy models command tells you which models your account can use, since the table in the documentation is shared across all packages. According to agy --help version 1.4.12, the models subcommand lists available models, and the --model flag takes a name from that list.
agy --version
agy models
On OPCAIOS's account on 03/11/2026, agy models returned 13 lines, each with a model code and a display name:
- Gemini 4 Argon: Low, Medium, High.
- Gemini 3.8 Flash: Low, Medium, High.
- Claude Opus 5.5: Low, Medium, High.
- Claude Sonnet 5.5: Low, Medium, High.
- GPT-OSS 120B: only Medium.
This is the list by account at time of check, may differ from your account. CLI splits each thinking level into a separate variant (code with suffix -low, -medium, -high), and no model here has xhigh or max variants.
What is thinking level (effort) in Antigravity?
Thinking level (effort) in Antigravity is the reasoning budget a model uses for each response: low level replies fast and light, high level thinks deeper and slower. The /docs/models page calls this thinking level and lists three levels Low, Medium, High in the app; the /docs/cli/commands/effort page (read on 03/11/2026) lists 5 levels the CLI recognizes, from light to heavy:
- low: fast response, suits quick lookup, syntax check, simple fix.
- medium: balance speed and quality for daily technical work.
- high: deep reasoning for complex architecture, hard bugs, refactor many files.
- xhigh: only with models that announce xhigh variant.
- max: highest budget, only with models that announce max variant.
Most adjustable models have low, medium, high; models with fixed thinking level have CLI report not adjustable. In the app's selector, Gemini 3.1 Pro only has Low and High. The description "suits work at" is from Google, not OPCAIOS measurement.
How to Change Models and Thinking Levels in the App and CLI?
Changing models in Antigravity has two ways: the model selector below the input field in the app (and IDE), or commands and flags in the agy CLI. Your choice in the app stays the same across messages within the same conversation.
In the Antigravity App
Open the selector below the input field, choose a model, then choose a thinking level; the selector also shows your remaining quota. If you change the model while an agent is running, that run still uses the old model (details in the FAQ section).
In the agy CLI
The /docs/cli/reference and /docs/settings pages (read on 03/11/2026) describe these ways:
/model: choose your preferred model, save it for later sessions (themodelkey in settings.json, saved by display name)./model <model> <prompt>: run one command on a different model without changing the default.--model: choose a model for this session only, don't save it./effort: open the thinking level slider;/effort low(or medium, high) change it right away and save as default.--effort: set the thinking level for one run, don't change the default.
agy --model="Gemini 3.8 Flash"
agy --effort low
agy -p "/effort" --output-format json
agy -p "Summarize the changes in the last commit" --effort low
In -p mode, /effort only shows information, it doesn't take parameters. Detailed CLI operations are in the post Antigravity CLI agy.
How Do Quotas and AI Credits Work in Antigravity?
Quotas in Antigravity are the amount of work an agent can do under your plan, while AI credits are extra purchases to use when you run out of quota, according to the /docs/plans page (read on 03/11/2026). The free plan has a quota that resets weekly; Google AI Pro resets every 5 hours until you hit the weekly limit; Google AI Ultra has the highest quota, also resetting every 5 hours. Google says quotas relate to how much work the agent does, complex work costs more, and quotas can be adjusted. The model selector shows two separate quota groups: the Gemini group and the Claude plus GPT group, each with a weekly limit and a 5-hour limit.
What you need to know when you run out of quota:
- Only Pro and Ultra plans can use purchased AI credits, calculated at Gemini Enterprise's standard consumption rates.
- The AI Credit Overages setting has two values: Never (wait for reset) and Always (use credits automatically).
- In the CLI, the
useG1Creditskey defaults to false; turn it on with/configor/settings. /creditsopens your credits balance;/usage(alias/quota) opens the quota table for each model.- Using your own API keys (BYOK) to raise quotas is not yet supported.
On pricing: the /pricing page on 03/11/2026 lists the individual plan at $0/month, with Google AI Pro and Ultra but no prices shown on the page, and the Gemini Enterprise plan from $30/person/month or pay-as-you-go. How to calculate the total cost of an AI tool is in the post cost of an AI tool replacing 3 staff.
When Should You Change Models in Antigravity?
Change models in Antigravity when you have a specific reason: your plan doesn't have that model, one quota group is exhausted, a model is about to be removed, or the work you're doing doesn't need deep thinking. The five cases below are based on documentation, not quality measurements:
- Model not in your plan: the free plan doesn't have Claude 5.5 or Gemini 4 Argon according to /docs/models; pick from what
agy modelsshows for you. - One quota group is exhausted: the Gemini group and Claude plus GPT group are separate, switching to a group with remaining quota lets you keep going without using credits.
- Model about to be removed: Claude 4.6 and GPT-OSS-120b are being removed on 18/11/2026, so switch your workflows before then.
- Light work like lookup, small fixes: Google describes low level as right for this kind of work; how much you save hasn't been verified.
- Hard work, errors repeat: try raising to high before switching model families.
An assumption, not a real person: someone doing OPC uses Gemini 3.8 Flash at medium level for daily work, runs /model <model> <prompt> to a Claude model when they need a second opinion, then checks /usage at the end of the day. New to model concepts? Read the post What is an LLM.
OPCAIOS Experience
OPCAIOS is the website opcaios.org run by one person using AI agents. These are true facts, verified on 03/11/2026:
- My main Mac has Antigravity app version 2.28.x, Antigravity IDE version 2.5.x and CLI
agyversion 1.4.12, used in parallel with Claude Code and Codex. How I divide roles among the three tools is in the post Antigravity, Claude Code and Codex in one system. agy modelson my account lists exactly 13 variants as mentioned above; there are no xhigh or max variants.- I have not compared quality, speed or limit consumption between models on the same task, so I don't recommend a fixed model.
- I have not measured productivity and have no numbers on time saved when switching models or thinking levels.
Frequently Asked Questions
Which model is best in Antigravity?
Antigravity documentation does not rank models, and OPCAIOS has not measured them so there is no "best" answer. The documentation only describes thinking levels: low for quick work, medium for everyday work, high for complex work. The practical way is to start with the model included in the medium-level package, and only switch when there is a specific reason like running out of quota or repeated wrong results.
Why don't I see Claude Opus 5.5 in the selector?
According to the /docs/models table read on 03/11/2026, Claude Sonnet 5.5 and Claude Opus 5.5 are not in the free package and Google AI Plus, and with Google AI Pro only apply to paid subscriptions, not trial versions. Gemini 4 Argon also requires the paid Pro package. Run agy models to see the real list for your account, as the per-account list may differ from the common table.
If I change the model mid-run, does the agent switch right away?
Not immediately. According to /docs/models, if you change the model while the agent is running, the agent continues using the old model until it completes the steps of the current turn or until you cancel. The next turn will use the new model, and that choice persists in the conversation. In the CLI, --model and --effort only apply to that session or run, while /model and /effort <level> save as defaults.
If I run out of quota does Antigravity auto-charge me?
Not automatically unless you enable it. According to /docs/plans, only Pro and Ultra packages can use purchased AI credits, and the AI Credit Overages setting decides whether to auto-use credits (Always) or wait for renewal (Never). In the CLI, useG1Credits defaults to false. Check your balance with /credits, your quota with /usage. This is documentation description, not financial advice.
Can I use a private API key to get more models or quota?
According to /docs/plans read on 03/11/2026, Antigravity does not yet support private keys (BYOK) to increase quota. Only the CLI has a modelProvider setting as "gemini" to send requests directly to Gemini API using your key, with customModelsConfig to add Gemini models to /model. To learn how to choose and manage AI tools with cost control, see OPCAIOS's AI application training course.
Sources: Google Antigravity documentation at antigravity.google: /docs/models, /docs/plans, /docs/cli/commands/effort, /docs/cli/credits, /docs/cli/commands/credits, /docs/cli/commands/usage, /docs/cli/reference, /docs/settings, /docs/faq and /pricing page; agy --help and agy models from Antigravity CLI 1.4.12; all read on 03/11/2026. Unverified: quality and speed between models, actual quota consumption per thinking level, pricing of Google AI Pro and Ultra packages; model list changes over time and by account. The experience is OPCAIOS's.
Translated from Vietnamese with AI. Read the original.
Want to apply this in your business?
Book a free 30-minute call (Vietnam time, in English or Vietnamese).
