Google Cloud has made Anthropic’s Claude Opus 5.5 and Claude Sonnet 5.5 available in Google Antigravity for Google Cloud customers, adding them to the same agent harness that already supports Google’s Gemini models. The post’s framing is that different tasks call for different models: Claude Opus 5.5 is built for complex work, while Claude Sonnet 5.5 is a faster, lower-cost complement. After an administrator enables the models, developers can select them directly from the model picker. Administrators can also select Google’s Gemini models, including Gemini 3.8 Flash and Gemini 4 Argon. The models apply across Antigravity 2.0, the Antigravity CLI, and the Antigravity IDE extensions for VS Code, JetBrains IntelliJ, and Visual Studio, all under a unified enterprise billing and security boundary. The release follows Google Cloud’s stated commitment to an open ecosystem, giving developers the right model for every job.

Having multiple model families inside the same agent harness unlocks practical multi-model workflows that improve code quality while controlling latency and cost. The first pattern is cross-model verification (‘LLM-as-a-judge’) for agentic loops and graph workflows. When the same model both writes and reviews its own output, model-specific blind spots can be reinforced. Pairing two distinct model families, one as implementer and another as validator, provides independent verification before the workflow moves to its next stage. In Antigravity you can run this pattern interactively or via automated subagents: plan and implement a multi-file change in planning mode with Claude Opus 5.5 or Gemini 4 Argon, then switch models, or invoke a one-shot /model check in the CLI, so a different model family critiques the implementation against your test suite and requirements.

The second pattern pairs reasoning models with Flash subagents. Complex tasks often combine a few high-reasoning decisions with dozens of high-volume mechanical lookups: searching files, grepping logs, checking lint output, and classifying test failures. Running every lookup on a frontier model can increase both latency and spend. Antigravity’s subagent customization system (.agents/agents/*.md) lets you assign model tiers (inherit, flash, or pro) to specialized subagents. Subagents configured with model: inherit use your active session model for deep reasoning subtasks; subagents configured with model: flash offload high-speed search, classification, and triage work to Gemini Flash models. The post gives two concrete files: a fast log-triage subagent in .agents/agents/test-triager.md, and a cross-model code reviewer in .agents/agents/gemini-judge.md. With that setup, you can delegate rapid test-and-classify loops to the triager and trigger the judge to validate the final diff.

Setup depends on the administrator path. For Gemini for Business administrators, third-party partner models do not draw from the pooled Gemini Enterprise Standard and Plus edition monthly token quota; usage is billed pay-as-you-go through the existing Google Cloud Billing account, so overages must be enabled. Configuration happens under Gemini Enterprise > Settings > AI developer tools in the Google Cloud Console, where you can set overage billing, spend limits, and authorize partner models. Administrators using Antigravity through the API in Gemini platform can enable Claude models in Model Garden and complete the standard Anthropic terms.

Developers update to the latest version of Antigravity 2.0, the Antigravity CLI, or the IDE extensions, then open the model picker or run /model. The active model can be switched between turns in the same conversation thread while preserving workspace state, open artifacts, and terminal context. One operational note: lightweight first-party Gemini models continue to run in the background for high-frequency client-side coordination such as context management and summaries; they consume first-party tokens and can incur charges. The full list of models supported in Antigravity across subscriptions is in the linked documentation.