Tharindu Indrajith

Human-Centered AI Scientist ||
CTO @ Pekoe Singapore

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK @GeminiApp @GoogleAIStudio we need Gemini 3.5 flash live speech to text API. Coz the Google Speech to text does not work well for strong accents. Gemini works extremely well with context and Google speech to text only works with local context.

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK @GoogleAIStudio Gemini in AI Studio consistently fails to deploy an app that uses Gemini. This should work extremely well, given how strongly it pushes to use Gemini-based services in the app, even though I don't need them. But it fails to implement. it

[Tharindu Indrajith @tikzoxs]

@GoogleAIStudio A Food & Nutrition Tracker and Planner for my Baby

[Tharindu Indrajith @tikzoxs]

@GoogleSmallBiz HR, Finance & Accounting. Period

[Tharindu Indrajith @tikzoxs]

@limatech_ar @madebygoogle They used to fix previously. Have been using only Pixels since 2017. When I send some issue via feedback usually it was fixed in the next month's update.

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK Please also fix the serving scaling issues of 3.1 inference. The model is always out of order in Antigravity, Gemini CLI, Firebase Studio. Even with Paid API keys from AI studio

[Tharindu Indrajith @tikzoxs]

@antigravity Can Workspace accounts also top up?

[Tharindu Indrajith @tikzoxs]

@sundarpichai Finally. Hope it's good

[Tharindu Indrajith @tikzoxs]

@tulseedoshi I think your release may have introduced a crazy bug to Gemini App reveling unrelated chats, system instructions, etc. not sure if there's a security or Privacy risk. Please fix it fast. Many people reported

[Tharindu Indrajith @tikzoxs]

@feelzyou Hello

[Tharindu Indrajith @tikzoxs]

@jocarrasqueira @antigravity @_anshulr @geekyouup 1) 3.1 is completely unstable in Antigravity and CLI. Only usable model is 3 flash in Gemini series.
2) Add BYOK like in Firebase

[Tharindu Indrajith @tikzoxs]

@patloeber Before that please fix bandwidth issues. 3.1 crashes all the time even with Paid API keys. Very slow. So many timeout issues.

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK @GeminiApp @Firebase @antigravity @GoogleAIStudio I get timeout errors, quota exceed errors and may other types of failures. Even in Firebase Studio with a Paid API key from AI studio. After 3.1 the experience has been extremely bad with errors and timeouts.

[Tharindu Indrajith @tikzoxs]

@pretty_sarlin Give in Sequence with like a 1 minute gap (assuming rat drinks like within 30 secs max). Since it takes exactly 10 hours, you will know which bottle killed based on the time of death. Only need 10 hours and 3 minutes.

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK It's with AI Pro. Even with AI Pro limits ran out after a few prompts for all models. Then reset in two weeks. Why not daily limits? What's with this 2 week reset time?

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK Before that please allow BYOK in Antigravity. Our team started using it but then have to wait 2 weeks for the quota to reset. Firebase Studio allows API keys. Please do the same to Antigravity

[Tharindu Indrajith @tikzoxs]

@GeminiApp

[Tharindu Indrajith @tikzoxs]

@OfficialLoganK please allow to select the default model. Now it's Fast by default. I quickly open Gemini and send a prompt only to realize it's going to fast. Then I have to copy the prompt refresh and run again in pro which is slower coz I have to refresh and re-run

[Tharindu Indrajith @tikzoxs]

@antigravity At least allow API key capability