Google Quietly Adds Anthropic's Claude Opus 5.5 and Sonnet 5.5 to Antigravity — but Only for Paying Users

Google has quietly added Anthropic's newest flagship models, Claude Opus 5.5 and Claude Sonnet 5.5, to the model picker of Antigravity, its agentic coding platform — placing a rival's best coding work directly inside the core of its own developer tool. The addition was spotted in Google's Antigravity model documentation on October 3, without any mention in the product's changelog.

Both models appear as "thinking" variants in the selector that sits beside Google's own Gemini models. Developers can now run Opus 5.5 or Sonnet 5.5 on their codebases without holding a separate Anthropic API key — but only if they are paying Google for it.

Antigravity is the autonomous coding workspace Google built after its $2.4 billion licensing and talent deal around Windsurf. It plans, writes, and tests code across complex software repositories, and it already carried older Claude models — which is exactly what makes the quiet update noteworthy. A company spending billions building its own frontier models has chosen to bake a competitor's newest releases into its flagship developer product.

The gap was real: Antigravity had been running Claude Opus 4.6 and Sonnet 4.6 even after Anthropic shipped Opus 5.5 on September 22 and Sonnet 5.5 on September 28. The documentation refresh brings both current models in.

The documentation notes that a reasoning-model selection persists within a conversation — swapping mid-run does not change the model used for the active turn.

Access is strictly tiered. Claude Opus 5.5 and Sonnet 5.5 are listed only for eligible Google AI Pro and Google AI Ultra subscribers, at roughly $19.99 a month for Pro and $99.99 or $199.99 for Ultra tiers. A trial Pro subscription does not unlock them, and users on Free, AI Plus, or Enterprise plans are shut out entirely.

The same documentation sets a hard date: Claude Opus 4.6, Claude Sonnet 4.6, and the open-weights GPT-OSS-120B model will be removed on November 2, 2026. Developers on non-paid tiers are already seeing an official notice in the Antigravity IDE stating that "Opus 5.5 is now available on paid Pro and Ultra plans" and that "third-party model access will no longer be available on your current plan starting on November 2, 2026."

After that date, free and trial-tier users will be restricted to Google's first-party Gemini models, while paid Pro and Ultra subscribers keep the full multi-vendor picker.

The practical difference between the two new models is speed versus depth. Sonnet 5.5 is positioned for faster code execution at lower token cost, suited to bugs, features, and documentation with a written spec. Opus 5.5 targets deep logic and long-horizon agentic workflows — the ambiguous, multi-step, security-sensitive jobs where a stronger reasoning model pays off.

For engineering teams, the eligibility matrix itself is now part of the test setup: a result produced with an eligible Claude 5.5 entitlement may not reproduce on a free, trial, or Enterprise account. With the November 2 retirement date for the older models also on the books, teams have about a month to audit their workflows and model selections.

The move underscores how multi-model coding tools have become: even as Google pushes Gemini across every plan tier, it is conceding that developers want the best rival models one picker away — for a price.