Ultra Plan: Unannounced 8x Increase in Third-Party Model Refresh Time — Requesting Immediate Clarification

Hello,

I’m an Ultra plan subscriber and I need to flag what appears to be a significant, unannounced downgrade to the service.

Until recently, all models in Antigravity — including third-party models such as Claude Sonnet 4.6 (Thinking), Claude Opus 4.6 (Thinking), and GPT-OSS 120B — shared the same ~5-hour refresh cycle under the Ultra plan. This was a major part of the value proposition that justified the Ultra subscription.

As of now, the refresh time for all third-party models has been increased to 1 day and 18 hours — approximately 42 hours, or 8x the previous cycle. Meanwhile, Gemini models remain unchanged at ~5 hours. I received no email, no in-app notification, and found no changelog entry about this change.

This has a direct and serious impact on my productivity. I use Claude Opus as my primary model for development work in Antigravity, and a 42-hour refresh window effectively makes it unusable for sustained work without burning through AI Credits — which defeats the purpose of paying for Ultra.

I’d like to understand:

  • Why was this change made, and why was it not communicated to Ultra subscribers?
  • Is this permanent, or is it a temporary adjustment due to capacity constraints?
  • Will Google restore the previous refresh cycle, or offer Ultra users an option to maintain their original quota levels?

I’ve been a committed user of this platform, but silently reducing service levels for paying subscribers is concerning. I hope this can be addressed promptly.

Thank you.

Best regards

5 Likes

just cancel suscribe…

4 Likes

If you are paying $250/month and using it primarily with Opus as your primary model, it is preferable to switch to Claude Max. I know Ultra offers great value for money, but server times, uff, have been a headache!

4 Likes