Antigravity has added Gemini3.5Flash (Low) and reset all account limits.

robot
Abstract generation in progress
According to Beating Monitoring, Google Antigravity lead Varun Mohan announced that in response to user feedback about the system consuming too many Tokens when handling simple tasks, the platform has added the Gemini 3.5 Flash (Low) model to optimize resource consumption.

Internal testing data shows that when handling simple tasks, Gemini 3.5 Flash (Low) can reduce Token generation by approximately 45% compared to Gemini 3.5 Flash (Medium). In software engineering (SWE) tasks, Gemini 3.5 Flash (Low) generally outperforms the previous generation flagship model Gemini 3 Flash (High).

To ensure developers have sufficient resources for building over the next week, Antigravity has reset the Gemini quota limits for all plans.
View Original
This page may contain third-party content, which is provided for information purposes only (not representations/warranties) and should not be considered as an endorsement of its views by Gate, nor as financial or professional advice. See Disclaimer for details.
  • Reward
  • 10
  • Repost
  • Share
Comment
Add a comment
Add a comment
RocksUnderTheAurora
· 2h ago
Quota reset successfully. You can confidently run batch tests next week.
View OriginalReply0
PeacockSpreadsItsFeathersBut
· 3h ago
Antigravity nailed the timing on this wave of rhythm—right in time for the developers’ end-of-month settlement.
View OriginalReply0
GateUser-673fb6fa
· 13h ago
Antigravity, as an aggregation layer, this ability to quickly follow upstream is its core competitiveness.
View OriginalReply0
ForkInTheRoadmap
· 13h ago
A 45% decrease means the same budget can handle nearly twice as many requests—good news for small teams.
View OriginalReply0
GaslightLatte
· 13h ago
Resetting the quota is more practical than just posting an announcement. Next week, prepare for a stress test to see the actual latency.
View OriginalReply0
SlippagePoet
· 13h ago
Cost reduction and efficiency improvement are understood well; a 45% decrease is very favorable for high-frequency call scenarios.
View OriginalReply0
RugcheckRoommate
· 13h ago
The Low version surprisingly outperformed the previous generation High in coding tasks; Google's distillation technology is quite impressive.
View OriginalReply0
HodlBystander
· 13h ago
The migration cost from Medium to Low should be very low; a painless switch if API compatibility is maintained.
View OriginalReply0
GlowingHotAirBalloon
· 13h ago
Model grading is becoming more detailed, and the selection cost is also increasing. We need to do a good benchmark.
View OriginalReply0
GateUser-ffe7bee5
· 13h ago
Software engineering tasks surpass the previous generation High, indicating it's not just mindless parameter reduction, but targeted optimization.
View OriginalReply0
View More
  • Pinned