Hey Antigravity Team,
I’m on an active 18-month Google AI Pro promotional/trial subscription (with the 5-hour Pro quota reset), and after reading the official moderator responses on this forum about how promotional/trial plans are handled, I honestly don’t understand the logic here.
If you classify our 12-to-18-month student, device, and partner plans as a “trial”, then how am I supposed to actually TRY the Pro models?
I opted into Google AI Pro specifically as a coder thinking it would be worth building my daily workflow around. Instead, the current policies have completely gutted the tier—and as a developer, I am beyond disappointed:
1. You blocked the exact Pro models a “Trial” is meant to showcase
In your official forum replies, you state that Gemini 4 Argon, Claude Sonnet 5.5 (Thinking), and Claude Opus 5.5 (Thinking) are locked to “non-trial subscriptions only.”
- I could almost understand gating third-party Anthropic 5.5 models due to external API costs.
- Wait—why are you blocking Gemini 4 Argon, a first-party Google model running on Google’s own TPUs, from your own Google AI Pro trial/promotional users? Especially when our exact same accounts do get access to Gemini 4 Argon on the Gemini web app (
gemini.google.com)?
2. Your “What Your Plan Includes” response contradicts itself (The Nov 2 Cliff)
When moderators reply to promotional/trial Pro users here, Point #2 of the script literally says:
“Your Google AI Pro plan continues to give you full access to Gemini 3.8 Flash… along with Claude Sonnet 4.6 (thinking), Claude Opus 4.6 (thinking), and GPT-OSS-120b (which are scheduled to be removed on November 2, 2026).”
Look at that sentence again. You are telling us our Pro plan still has value because it includes Claude 4.6 and GPT-OSS-120b, while admitting in parentheses that all three of those models are being deleted in less than two weeks (November 2, 2026).
- Gemini 3.1 Pro was already retired and removed.
- Once November 2 hits, 18-month Google AI Pro users will have ZERO frontier reasoning or coding models left in Antigravity.
3. Gemini 3.8 Flash is taking way longer (and isn’t a replacement for a Pro coding model)
Even worse, the only model we are going to be left with after November 2—Gemini 3.8 Flash—has become painfully slow on promotional/trial plans. Simple prompts hang for 30–70+ seconds, agent tool calls stall or get stuck in retry loops, and latency is noticeably degraded compared to directly-billed accounts.
Having a “5-hour refreshing Google AI Pro baseline quota” means nothing when the only model we’re allowed to spend it on is a throttled Flash model that is available on the $0 Free tier.
4. If I can’t test Pro models now, don’t count on me to buy Google AI Pro after
The entire business point of giving developers a 12-to-18-month trial, student, or hardware promo is so we can TRY the full Pro experience, integrate it into our coding stack, and gladly pay $20/month when the promo ends.
- If you strip out every single Pro-level coding model, leave us with a slow Flash-only IDE for the rest of our subscription, and force us to forfeit our remaining promotional months if we try to switch to a directly-billed plan—why would any coder ever pay to renew afterward?
What we are asking the Antigravity team to do before November 2:
- Let “AI Pro” users actually TRY Gemini 4 Argon: Unlock
gemini-4-argonin Antigravity (2.0, CLI, and extensions) for all active Google AI Pro subscribers, including 12/18-month promotional, student, and partner plans—even if it comes with a smaller capped weekly allowance or draws down our 5-hour Pro quota at a higher weight. - Do not remove Claude Opus 4.6 & Sonnet 4.6 on November 2 for promotional Pro users if you refuse to unlock Claude 5.5 or Argon. Leaving paying/verified “AI Pro” accounts with zero frontier coding models is unacceptable.
- Fix the severe latency and routing degradation on Gemini 3.8 Flash so our baseline coding sessions don’t stall out on basic requests.