Hi Antigravity team,
Requesting that Claude Opus 5.5 be added to the model picker, and sharing why I think it makes sense for Google as well as for users.
1. Lower cost per task. Opus 5.5 is priced at $4/$20 per million tokens versus $5/$25 for Opus 4.6 — a 20% cut on both input and output. Since Antigravity serves Claude through the Vertex integration, a cheaper and more capable model should reduce serving cost per task, not increase it.
2. Better fit for Antigravity’s core use case. Opus 5.5 is specifically built for long-running agentic coding, which is exactly what Antigravity agents do. The current 4.6 thinking models are reported to burn quota roughly 4x faster than Gemini because hidden thinking tokens count against limits, and they have hit frequent 503 capacity errors.
3. The catalog is four generations behind. Anthropic has shipped Opus 4.7, 4.8, 5, and now 5.5, but Antigravity has been on 4.6 since February. Every other major coding tool (Claude Code, Cursor, Zed) offers the current generation. For paying Ultra subscribers this is the single biggest reason to consider switching platforms.
4. Volume can be managed with quotas, not with stale models. If the concern is total spend or steering usage toward Gemini, that is achievable with existing quota controls while still offering the current model generation to those who need it. A generation-old model with tight quotas is the worst of both worlds for serious users.
I understand Gemini remains the primary model and this would be a supplementary option. But for users whose work depends on current-generation Claude, either adding Opus 5.5 (even with strict quotas) or supporting bring-your-own-key would make Antigravity a viable primary tool instead of a fallback.
Happy to provide any additional details. Thanks for the great platform and for list
ening to feedback.