Artificial intelligence news.
Straight from the original announcements.

Claude Opus 5.5 and Sonnet 5.5 arrive in Google Antigravity

October 20, 2026, 18:00Source: Google Cloud
Claude Opus 5.5 and Sonnet 5.5 arrive in Google Antigravity
Image: Google Cloud

Google Cloud has made Claude Opus 5.5 and Claude Sonnet 5.5 available in Google Antigravity for its customers. Once an administrator enables them, developers can pick either model from the model picker alongside Google's own Gemini models.

Key points
  • Claude Opus 5.5 and Claude Sonnet 5.5 are now available in Google Antigravity for Google Cloud customers.
  • Administrators must enable the models, after which developers choose them in the model picker across Antigravity 2.0, the CLI and IDE extensions.
  • Partner model usage is billed pay-as-you-go via Google Cloud Billing, not from the pooled Gemini Enterprise token quota.
  • Google describes cross-model verification and Flash subagent pairing as two ways to combine models.

Two Claude models, one agent environment

According to Google Cloud, Claude Opus 5.5 is built for complex work, while Claude Sonnet 5.5 is a faster, lower-cost complement. The company says the addition reflects its commitment to an open ecosystem and to giving developers the right model for every job, whether that is a quick bug fix or a complex migration.

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 recently released Antigravity IDE extensions for VS Code, JetBrains IntelliJ and Visual Studio. Google Cloud says this happens under a unified enterprise billing and security boundary.

Developers can switch the active model between turns in the same conversation thread, and workspace state, open artifacts and terminal context are preserved, according to the post.

Two multi-model engineering patterns

Google Cloud's post describes two workflows that rely on having several model families inside the same agent harness.

Cross-model verification. The post argues that using one model to both write and review code can reinforce model-specific blind spots. Instead, one model family acts as implementer and another as validator, or judge. In practice, a developer can start in planning mode with Claude Opus 5.5 or Gemini 4 Argon to design and implement a multi-file change. Before committing or advancing the loop, they switch models, or run a one-shot /model check in the CLI, so a different family critiques the work against the test suite and requirements.

Reasoning models with Flash subagents. Complex tasks often mix a few high-reasoning decisions with many mechanical lookups, such as searching files, grepping logs, checking lint output or classifying test failures. Running all of these on a frontier model can raise latency and spend, the post says. Antigravity's subagent customization system, defined in .agents/agents/*.md files, lets users assign a model tier of inherit, flash or pro to each subagent. Subagents set to inherit use the active session model, while those set to flash send search, classification and triage work to Gemini Flash models.

The post gives two examples: a test-triager subagent on the flash tier that runs unit tests and classifies failures, and a cross-model-judge subagent on the pro tier that reviews a workspace diff and returns a pass or fail verdict with line references.

Setup and billing for administrators

Google Cloud separates the steps by how a team uses Antigravity.

  • Gemini for business: 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. Administrators configure overage billing, spend limits and partner model access under Gemini Enterprise, Settings, AI developer tools in the Google Cloud Console.
  • Gemini platform: Administrators using Antigravity through the API can enable Claude models in Model Garden and complete the standard Anthropic terms.
  • Developers: Update to the latest Antigravity 2.0, CLI or IDE extensions, then open the model picker or run /model.

Google also notes that lightweight first-party Gemini models keep running in the background for high-frequency client-side coordination such as context management and summaries. These consume first-party tokens and can incur charges.

What it means for teams

For Google Cloud customers, the change adds Anthropic models to a coding environment that previously centered on Gemini, without leaving the same billing and security boundary. Teams must have an administrator enable the models and turn on pay-as-you-go overages, since partner model usage is not covered by the pooled Gemini token quota. Developers who want to try the verification or subagent patterns can start from the configuration examples Google Cloud published, and the full list of supported models by subscription is in the Antigravity documentation.

Related news

Claude Code Session Costs: Opus 5.5 Median $0.31, Sonnet 5.5 $0.16
For developersOctober 30, 2026

Claude Code Session Costs: Opus 5.5 Median $0.31, Sonnet 5.5 $0.16

Claude Code is free to install, but the model it calls is not. According to OpenRouter, a short session of 2 to 9 turns on an API key costs a median $0.31 with Claude…

Amazon recaps Q3 2026: new Bedrock models, TorchNeuron, AWS Continuum and more
For developersOctober 26, 2026, 14:00

Amazon recaps Q3 2026: new Bedrock models, TorchNeuron, AWS Continuum and more

Ahead of its third-quarter earnings report on October 29, Amazon has published a roundup of its biggest announcements since the last results, led by new frontier models…

Google Cloud adds Anthropic-managed option for running Claude models
Products and toolsOctober 22, 2026, 23:00

Google Cloud adds Anthropic-managed option for running Claude models

Google Cloud has made an Anthropic-managed deployment option for Claude models generally available, giving customers a second way to run Claude next to the existing…