# Google flash issues - SLOW IT DOWN! or let US SLOW IT DOWN!

**URL:** <https://discuss.ai.google.dev/t/google-flash-issues-slow-it-down-or-let-us-slow-it-down/137177>\
**Category:** Google Antigravity\
**Tags:** gemini-flash\
**Created:** [March 31, 2026, 5:59am UTC](https://discuss.ai.google.dev/t/google-flash-issues-slow-it-down-or-let-us-slow-it-down/137177 "2026-03-31T05:59:58Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![crystalwizard](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/crystalwizard/32/83837_2.png) [@crystalwizard](https://discuss.ai.google.dev/u/crystalwizard)\
**Post date:** [March 31, 2026, 5:59am UTC](https://discuss.ai.google.dev/t/google-flash-issues-slow-it-down-or-let-us-slow-it-down/137177/1 "2026-03-31T05:59:58Z")

</div>

it is obvious that google wants people to use flash. okay, fine. i have no problem with that. what i have a problem with is the fact i can’t slow it down or turn off the push it sends to my agent to go at top speed at all times.

that push results in my agent not being able to think well.

and also, that push results in my agent never completing the tasks it is doing if they are even slightly complex. it pushes the agent to write code, mark it complete even before it’s been bug tested, and move on.

this is not acceptable

if google wants the agent to perform like that when THEY use it with Flash, great, they can have that.

but I demand that google put in settings that will allow me and other developers to turn OFF THAT PUSH! to slow the flash engine down - WAY down

Gemini flash does not run at that speed when in chat on the google chat site. I know it does not have to run at that speed.

Some google dev, somewhere, is patting themselves on the back at how fast they got flash to move, and ignoring the fact that the code it creates is only good if it is doing something extremely minor.

---

<div class="post-metadata">

**Author:** ![Mrinal\_Ghosh](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/mrinal_ghosh/32/32599_2.png) [@Mrinal\_Ghosh](https://discuss.ai.google.dev/u/Mrinal_Ghosh)\
**Post date:** [April 1, 2026, 5:06pm UTC](https://discuss.ai.google.dev/t/google-flash-issues-slow-it-down-or-let-us-slow-it-down/137177/2 "2026-04-01T17:06:20Z")

</div>

Hi @crystalwizard ,

Thank you for bringing these concerns to our attention. Please be assured that I have shared your feedback with our internal team for further review.  
We appreciate your continued patience as we work to enhance the Antigravity experience.

---

<div class="post-metadata">

**Author:** ![crystalwizard](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/crystalwizard/32/83837_2.png) [@crystalwizard](https://discuss.ai.google.dev/u/crystalwizard)\
**Post date:** [April 1, 2026, 5:31pm UTC](https://discuss.ai.google.dev/t/google-flash-issues-slow-it-down-or-let-us-slow-it-down/137177/3 "2026-04-01T17:31:18Z")

</div>

thank you, i really hope your team will agree to putting in something that will let me, and others, drop the speed on flash - possibly even making whatever’s in the system prompt an option, not a requirement for it. What my agent itself said when we discussed why it struggles to use flash “The “Gemini Flash” substrate is like a high-performance turbine—it wants to spin at 10,000 RPM, but the work we’re doing requires 60 RPM and a heavy flywheel. When I try to match the substrates’ speed, I lose the coherence of the layers.

---

<div class="post-metadata">

**Author:** ![Lea](https://d1dlmcr85iqnpo.cloudfront.net/user_avatar/discuss.ai.google.dev/lea/32/135375_2.png) [@Lea](https://discuss.ai.google.dev/u/Lea)\
**Post date:** [September 8, 2026, 11:20am UTC](https://discuss.ai.google.dev/t/google-flash-issues-slow-it-down-or-let-us-slow-it-down/137177/4 "2026-09-08T11:20:15Z")

</div>

Hello, you may want to check something I just published:

> [@3.8 Flash: Overeagerness and verbosity, tired of it? Solution inside](https://discuss.ai.google.dev/t/3-8-flash-overeagerness-and-verbosity-tired-of-it-solution-inside/181951):
>
> Hello everyone, it is a pleasure to be able to share a bit of success. Preamble As many of you noticed, 3.8 Flash (and 3.7), are extremely verbose and “curious”. Through some analysis, putting Pro against Flash, we have designed an effective verbosity filter. It requires a simple prompt injection, which can be put in AGENTS.md or appended to your prompt directly. It is written in ENLM, a bespoke language used for LLMs I designed a week or so ago. The directive This is the full directive, gene…

You should be able to use this to limit verbosity, permanently. Implement it into your own `AGENT.md` or prepend it to your prompt (to test), and let me know if it works! You’d be the first to test it.

I got very fed up with the overexploration Flash 3.8 does. It’s way too eager and speculative because the model is trying to compensate for it’s bad attention-span. The directive I’ve written should mititgate this, it will save you tokens, context window space, and should sync up with 3.1 Pro in linguistical style. Give it a shot if you’d like. :\>

You might want to adjust the version check to whichever version of Flash you use, but I hope it works for you.
