I don’t understand why we still have zero support for GPT Models in Antigravity.
While other IDEs and agentic platforms are day-zero-integrating the latest OpenAI models, Antigravity feels like it’s stuck in a bubble. We are trying to build complex autonomous systems, and limiting us to Gemini 3 and Claude is a massive bottleneck.
Can someone explain why we are stuck with these bottom-tier Claude limits in Antigravity?
Even in the Pro plan, the current Claude integration is not useful at all. The limits are so restrictive that the “agentic” part of this IDE becomes a joke the second you try to do serious work.
We need GPT 5.3 Codex. It is the only model that actually handles the reasoning needed for complex tasks without hitting a wall every five minutes. The fact that the Antigravity team hasn’t prioritized this integration—while simultaneously gimping the Claude options we DO have—is a massive failure. Stop being a Google-only echo chamber and give us the tools that actually work.
Is Antigravity an open platform for developers, or is it just a vessel for Google models? If you don’t bring in the 5.3 Codex engine soon, there is no reason for serious engineers to stick with this IDE. Address this now?
2 Likes
That’s true regarding the company relationships, but my point isn’t about who owns or funds whom. It’s about capability and ecosystem alignment.
If Antigravity aims to stay competitive as a serious development platform, then support for the most advanced coding models matters—regardless of whether they come from OpenAI, Google, or Anthropic.
Right now, models like GPT-5.3 Codex (from OpenAI), Gemini (by Google), and Claude (from Anthropic) each have different strengths.
The question isn’t who competes with whom — it’s whether Antigravity wants to support the strongest tools available for its users.
If developers are increasingly relying on GPT-based coding workflows, lack of support becomes a practical limitation, not a political one.
That’s the core issue I’m raising.
Agreed, but Codex is an extension so who ever owns the extension needs to update it, that said better if it was baked into Antigravity. Hopefully Google Ships another model soon because Gemini 3 Pro, is def WAY behind for coding.
Actually the reason is very very clear the hardware is the bottle neck, Anthroipic can work on TPU and this is how Google can afford such in the package and I understand this very well from hardware prespective, however I totally agree including other models, I think its not Google’s call in this case its OpenAi adoption of the TPU which I dont think it will be very soon due to the chip game