So Google really needs to add Sonnet 5 to Antigravity, as it’s speculated to release today, and since they skipped Opus 4.8 and 4.7, I hope they don’t skip this model, as we will be able to get a higher rate limit from it and another option than Google models that’s actually competitive and will give its own insight, and since Google probably didn’t add Opus models due to cost, I hope they add this efficient model, as it’s easier and cheaper to serve.
I completely agree. I hope they add this model quickly.
Sonnet 5 Medium cost is great. Would be a good addition to Antigravity.
If they don’t add Sonnet 5 to Antigravity, then with almost 100% certainty I’ll cancel my subscription and switch to some more advanced platform that adds new models on the day they’re released, like GitHub Copilot. Like, what is this nonsense? Are you making a normal product that people will actually use, or are you just playing games? Why don’t you have all the models that people need?
This is just a way for Google to save money, but they don’t realize the effects it’s having on developers. And yet, the sad reality is that they still probably won’t add it, as they did with Claude Opus models. It was more of a promotional period kind of thing, They just don’t update the models. However, it’s getting really frustrating now, and it’s best to just switch subscriptions at this point if they don’t soon update the catalog. As models are more important than the agent harness, this seems like a big disappointment. I had high hopes for this IDE, considering it’s the only one that gave a free tier for models, but now, even with a subscription, it doesn’t seem that appealing. Personally speaking, I don’t think this is fair and is an issue that needs to be immediately fixed, as almost all IDEs and CLIs have already integrated these models. So, I’m still waiting, and they keep telling us to use Claude Code or our own API keys, as when I asked them on YouTube. But then, what’s the point of Antigravity? Why not just use VS Code? It seems like a useless product without these integrations, and they don’t realize how many customers they are losing because of this. I hope a Google employee sees this thread and fixes this issue.
I don’t have high hopes, unfortunately, as people have been asking for Opus’ latest models for a while now, and they have just been ignoring us. But I believe Sonnet 5 is an essential model because we can get more usage and a smarter model.
I agree and hope that the Antigravity team finally listens to us and hopefully also adds the Opus models alongside Gemini 3.5 Pro, or even earlier than that, as we are missing out on some serious quality. And I personally don’t think that Antigravity is facing problems integrating the model with the system prompts and tool instructions, but rather that it’s a financial decision to save money—typical corporate greed—but I still hope for it.
In a few days GPT 5.6 SOL will be released to the public. In Codex even with GPT 5.5-Extra-High, with a 23 euros subscription I can get done 15x more tasks than I can do in antigravity with an equal price subscription (google AI pro tier). How is it possible that using GPT 5.5-Extra-High I can do for same price so much more work? Surely google can afford to allow higher rate limits to users, there is no way that limits should be so low for someone paying 22 euros monthly. The pro model burns too much compared to competition and it’s got to the point that it’s not just about having less smart models, but it’s about what google is offering in terms of rate limits.
I completely agree, and I guess it’s time to move towards Codex now; even though I really like that they also had a free tier, but now that I’m going to have to buy a subscription, I guess I’ll just buy Codex as it has better rate limits and better models. However, if Antigravity adds the newer models to it, then I would love to switch back, but I guess until then it will be a sad goodbye, as it could’ve been a good product just because they tried to cut budgets and focused on multiple projects at the same time, they couldn’t focus on making one really good, like Claude Code and Codex are.
I wonder when this will happen. My time developing in Antigravity is gradually decreasing, since I find myself using Sonnet 5 and GPT 5.6 more and more than the agents available in the IDE. #sad 
I still hope that they add gemini 3.5 pro with the upcoming opus 5 and sonnet 5 this just shows how behind they are in terms of models
There is still no update on the model catalogue. Gemini 3.5 Pro is nowhere to be seen. Even if Google releases a model that’s capable of beating upcoming Opus 5, Sonnet 5, or even Fable 5, they’re gonna spend another six months to develop Gemini 3.6 Pro. By then, Claude will have an even more impressive model catalogue, and Kimi K3 is most likely better than Gemini 3.5 Pro. Therefore, adding Claude models to the IDE is the only way to save it, as there’s no future otherwise, and no feature can save this harness. There is also speculation that Gemini 3.6 or 4 Flash may come before, meaning Antigravity is gonna be based on speed rather than intelligence. This is because they don’t add Claude models for intelligence and can’t compete with other AI labs for an intelligent model. So, I guess it’s time to leave this harness, and it’s been a good run until Antigravity 2.0.