I want to add a more data-grounded voice to the several threads already asking about this. I’m not here to vent — I’m here to make the case for why this matters competitively and what the team could do about it structurally.
The facts
Anthropic released Claude Opus 4.7 on April 16, 2026. It has now been over 24 hours.
The model is already available on: Claude Code, Cursor, Amazon Bedrock, Google Cloud Vertex AI, Microsoft Foundry, and GitHub Copilot.
As of right now (April 17), Antigravity’s model picker still caps at Claude Opus 4.6 (Thinking).
Why this specific model matters
This isn’t an incremental point release. The benchmark jumps are the largest within the Opus family:
Benchmark
Opus 4.6
Opus 4.7
Delta
SWE-bench Pro
53.4%
64.3%
+10.9 pts
Computer-use
54.5%
98.5%
+44 pts
Vision resolution
~1.25 MP
3.75 MP
3×
CodeRabbit bug detection
55/100
68/100
+24%
For developers doing serious agentic coding — multi-file refactors, long-running autonomous tasks, vision-guided workflows — this is a generational leap, not a minor upgrade.
The competitive reality
Antigravity positions itself as a premium, model-agnostic agentic coding platform. Its direct competitors acted within hours:
Cursor — Opus 4.7 available in model picker on release day.
Claude Code — Ships with it natively (it’s Anthropic’s own tool).
GitHub Copilot — Available via Bedrock/Vertex integration same day.
Amazon Bedrock — GA on release day.
Meanwhile, Antigravity users — including paying Ultra subscribers — are still on 4.6. Every hour this persists, developers who rely on best-available models are running their critical workloads in a competing tool instead of here.
This happened before
The same lag occurred with Opus 4.6. Community threads appeared, the team acknowledged, and it eventually landed. But the pattern is becoming predictable:
Major model drops.
Competitors onboard it in < 12 hours.
Antigravity community posts escalate.
Team files internal request.
Days/weeks pass.
This cycle erodes trust. Users choosing Antigravity for its multi-model flexibility expect that flexibility to be responsive.
What I’m asking for
Ship Claude Opus 4.7 in the Antigravity model selector ASAP. The model is already on Vertex AI — Google’s own infrastructure. The integration path is short.
Establish a public SLA for major third-party model onboarding. Even a stated goal of “within 48 hours of GA on Vertex AI” would set expectations and show the team takes model freshness seriously.
Communicate proactively. When a major model drops, a simple pinned post saying “We’re aware — Opus 4.7 ETA is X” would prevent five separate frustrated threads.
Bottom line
The model is on Vertex AI — Google’s own cloud. The integration path is short. Every hour without it is an hour where users who need the best available model are running their workloads somewhere else.
I am seriously wondering if they have some algorithm that tells them they don’t have to fix their server issues or the wrapper issues or the GUI issues, and x% of users won’t cancel, then another one that tells them that only x% of users will cancel and switch to a platform with the Opus 4.7 model, and the rest will stay and complain for 2 months before they have to do anything.
I know how this may sound, but a multi-trillion dollar company absolutely has to have some awareness of these things, and by the looks of the community here you would think Google has no clue.
I honestly can’t figure out why they don’t just take a few high quality devs and have them fix antigravity. It would take a actually good team a day. Then they could upgrade their servers they use.
Meanwhile they are probably working on something to replace AG, that will be just as buggy and advertised as AG, and the masses won’t notice or bother posting about it until the next model is about to release.
These huge companies need to have some accountability and actually release non beta software before they take 99% of their resources and start new projects.
I agree, wishing for the latest is pointless because if we had it we wouldn’t be able to use it. It’s obvious whomever was behind this project, since there have been NO updates (like program updates or updates on the situation…) despite this going on for the entire work week now. Now that the weekend is here, I expect this to last for another few days which is absolutely. When Opus 4.6 was released, we had it in Antigravity the same day so it’s super apparent that the team or individual behind Antigravity was removed or restructured because it was probably costing Google too much (hence the new very restrictive limits and errors). I’m likely going to downgrade my plan and shift over to Claude Code, while that platform isn’t perfect, it atleast is transparent as to how much hourly and weekly usage you have remaining, and how much chat context has been consumed.
I expect to see Antigravity listed on the Google website by next year, if not sooner.
There is something seriously wrong at google when it comes to making products. It’s a long running issue, and it’s unfortunate, but it appears to mean they are incapable of properly supporting any product they release. At least incapable of meeting the expectations which have been set by other companies that release commercial products. Antigravity was so cool that I was willing to overlook this, but I shouldn’t have. And now I’m paying for it, like all of you. I guess they can just continue to feed off the money printing machine they are until they finally go down like the titanic, but in the meantime, we all have to remember never to engage with a google product unless we are willing to treat it like a broken toy. And even that is probably not a good idea. I wish a tiny startup had made this product, because at least then they would support it. Probably even a one person dev could do a better job.
Concordo plenamente. Depois de pagar 2 meses pelo plano Ultra, que só assinei para liberar o limite do Antigravity, vi que depois de 20 e poucos dias de uso, os limites a cada dia diminuíam, até o cúmulo de eu usar 1 prompt, para uma correção boba no sistema, e o limite do modelo acabar naquele dia. Essa semana cancelei o plano Ultra e se der, vou cancelar todos os serviços que tenho com o Google, pois cansei de ser tratado como lixo por essa empresa.
Fiz o mesmo! Cancelei o plano Ultra exatamente por esse mesmo motivo. Falta total de transparência dessa empresa chamado Google, que espero que venha a falir nos próximos anos e que seja substituída por empresas que respeitem o cliente. Falta total de respeito, falta total de transparência.
+1 as a small startup, we spend on 5 ultra accounts. Not having latest models impacts our ability to compete, we’ll have to move if Google can’t keep up.
Why would we need opus 4.7 when we dont have quota enough to use Opus 4.6 more then 2-3 prompts a week. Also, I think the real reason opus was there at al in the beginning was that they needed something that worked to teach gemini 3.0 with feedback data
you all are complaining about a ide or the only ide that is providing so much for free
just think about the comparasion you all are making you all are comparing a free ide with top tier ai agents with expensive ides with just slightly better ai agents .