# Gemini 3.7 Flash live **URL:** \ **Category:** Announcements\ **Created:** [September 17, 2026, 4:08pm UTC](https://forum.aiendurance.com/t/gemini-3-7-flash-live/212 "2026-09-17T16:08:37Z")\ **Posts on this page:** 3\ **Page:** 1
**Author:** ![MarkusRummel](https://yyz1.discourse-cdn.com/flex003/user_avatar/forum.aiendurance.com/markusrummel/32/8_2.png) [@MarkusRummel](https://forum.aiendurance.com/u/MarkusRummel)\ **Post date:** [September 17, 2026, 4:08pm UTC](https://forum.aiendurance.com/t/gemini-3-7-flash-live/212/1 "2026-09-17T16:08:37Z")
Gemini 3.7 Flash is now the default chat model, replacing 3.0 Flash. It scores clearly higher than its predecessors in Google’s benchmarks, and in our own evaluation it made fewer factual slips and was about twice as fast to respond. Let me know if there’s any issues or bugs please. ![image](https://canada1.discourse-cdn.com/flex003/uploads/aiendurance/original/1X/e47eb1dab34df8354503957a25148f427e976b3f.png) ---
**Author:** ![R\_Handle](https://avatars.discourse-cdn.com/v4/letter/r/57b2e6/32.png) [@R\_Handle](https://forum.aiendurance.com/u/R_Handle)\ **Post date:** [October 3, 2026, 3:53pm UTC](https://forum.aiendurance.com/t/gemini-3-7-flash-live/212/2 "2026-10-03T15:53:20Z")
I used flash for other purposes, it’s quite impressive, especially since 3.5. I tried 3.5 Flash-lite the other day to capture data for an agent, faster than gemini 3.8, did the job just as well, and cost significantly less. ---
**Author:** ![MarkusRummel](https://yyz1.discourse-cdn.com/flex003/user_avatar/forum.aiendurance.com/markusrummel/32/8_2.png) [@MarkusRummel](https://forum.aiendurance.com/u/MarkusRummel)\ **Post date:** [October 4, 2026, 8:37pm UTC](https://forum.aiendurance.com/t/gemini-3-7-flash-live/212/3 "2026-10-04T20:37:27Z")
Glad you like it. Evaluated Flash 3.8 as well but performance was similar to Flash 3.7 and 3.7 was twice as fast