Gemini 3.6 Flash - will be Leaving the Google Antigravity Soon

Google Antigravity
, ,
1

Actually if Gemini 3.6 Flash leaves the Antigravity anytime, that would be the end of my journey on Google’s Antigravity :joy:

3.6 is the only BALANCED model I find in antigravity. Models like 3.8 are on another level, but quota usage is very high compared to 3.6 and 3.7.

image
image703×374 24.4 KB

2

Hi @SuneraX,

Thank you for sharing this feedback! As noted by the “Leaving Soon” tag in the model picker, we are transitioning toward Gemini 3.8 Flash as our primary Flash model, though we have shared your feedback about Gemini 3.6 Flash with the product team.

Regarding quota usage: Gemini 3.8 Flash actually has the exact same per-token quota rate as Gemini 3.6 Flash and 3.7 Flash and draws from the same Flash quota pool.

If you are noticing higher quota consumption with Gemini 3.8 Flash, it is likely because Gemini 3.8 Flash (High) uses an uncapped thinking budget and performs deeper multi-step reasoning and tool calls per turn. To get the same balanced quota usage you had with Gemini 3.6 Flash while keeping the quality improvements of 3.8:

  • Switch the model picker from Gemini 3.8 Flash (High) to Gemini 3.8 Flash (Medium) or Gemini 3.8 Flash (Low), which cap thinking tokens per turn to a fixed budget.
  • You can also set verbosity to medium/low in /config > Verbosity (if using the CLI) and start a fresh conversation when switching between unrelated tasks so large historical context isn’t re-sent on every turn.