Google Gemini

We haven't received many problem reports for Google Gemini yet.

If Google Gemini is down for you, you might be one of the first to encounter an outage - please consider checking back later. It's also possible that the issue is local or related to your internal network or system.

Google Gemini reports

Google Gemini is an AI-chatbot. When Google Gemini is down, the service may be unable to repond to prompts, the API may be down or login isn't working.

Reports about problems with Google Gemini

cryptopunk7213

05/13/2025, 03:36 AM

Time & skill needed to code apps trending to zero. Time that agents can spend working on hard problems 2X-Ing every 7 months Sign ups for Gemini, Claude & openAI access going through the roof. China closing the gap on Chip manufacturing Personalised agents becoming people’s

TarouLovesYou

05/13/2025, 03:03 AM

Has anyone experienced the supposed drop in performance with Gemini 2.5 pro? Is it now fixed?

JustVibesFyi

05/13/2025, 02:19 AM

Day 6 and 7 of posting daily updates: Didn't post an update yesterday as it was quite a busy day plagued by technical issues. stealth: Work on a major feature is underway agent: Added gpt and gemini for different tasks - using vision too. The workflow is almost complete main:

JustVibesFyi

05/13/2025, 02:17 AM

tom_doerr

05/12/2025, 11:49 PM

I prompted Gemini 2.5 Pro to refactor my app, which kept on breaking in weird ways with every edit. Can't believe it now works flawlessly, with no need to fix issues or restore lost features, even after prompting it to keep refactoring multiple times

arafatkatze

05/12/2025, 11:12 PM

@otakakr Sorry for all the trouble, hopefully gemini improves soon and then you can have a better experience. For now does sonnet do the job for you?

ziakarim

05/12/2025, 09:00 PM

Kind of amazed at the number of times Gemini 2.V solves intricate and subtle issues that are completely lost on both GPT and Claude.

LuttrellAI

05/12/2025, 08:16 PM

@arafatkatze @cline FWIW: I haven't had any issues. Had been using Gemini because Claude was hit and miss actually. (There is occasional stupidity but that might be all me.)

jayrodge15

05/12/2025, 08:13 PM

@joshwoodward @GeminiApp Gemini live with camera/screenshare on iOS when?

nickwm

05/12/2025, 06:24 PM

@K_to_Macro This is an issue with the model and not Cursor. We’ve been working closely with the Gemini team to help them resolve it. They released an update earlier this month that we’re now serving which should improve things.

itetnaa

05/12/2025, 03:14 PM

@theo I am actually about to crashout why tf is Gemini 2.5 Pro not working. I have been clicking on the retry button like a thousand times. ITS STILL NOT WORKING.

Adam Landeskog

05/12/2025, 09:52 AM

I just get the error "something went wrong" regardless of the prompt I put in, it can be anything, really

isala404

05/12/2025, 09:46 AM

Is it me or gemini in ai studio is not working at all?

Abdullah Alhamidi

05/12/2025, 09:39 AM

Currently down for me, the requested input keeps getting returned and the response hangs after a few seconds/minute.

MaxLikang

05/12/2025, 09:07 AM

My AI Gemini 2.5 Pro is not working.

MaxLikang

05/12/2025, 09:05 AM

@_yamanori My Gemini AI 2.5 Pro is not working.

HCSolakoglu

05/12/2025, 09:04 AM

Oddly, in the Gemini app, G2.5Pro keeps giving errors. This started with the latest update on May 6th, which was also when a new model update was released. Is another model update, or maybe Ultra, coming soon?

edwarddegil

05/12/2025, 07:33 AM

@OfficialLoganK @v0 even though I haven’t used the 2.5 pro in the gemini app at all today, it’s giving me an error whenever I try to send a message. please resolve this issue.

KuittinenPetri

05/12/2025, 07:13 AM

@Teknium1 Is it just me, but Gemini 2.5 Pro and Flash became worse in everything else than coding? The safety settings even when set to block NONE end up "Sorry, I cannot help you Dave". PS. And you are correct, the API can often refuse connections. I haven't seen such error rate with

iamioririo

05/12/2025, 02:09 AM

@slow_developer I’ve experience similar over censoring issues with Gemini until explicitly setting each filter to block none

preshenhq2_0

05/11/2025, 06:33 PM

@hourIyhoroscope Leo, we know you had problems with Gemini, recovery should work. #GranHermano #GrandeFratello #zeudiners #helevier #zelena #yulioli #zeuchi #tommavi #bbvipal4 #heleners #uzaksehir #shailenzo

SirTHobbes

05/11/2025, 10:13 AM

@tom_doerr code with Claude, fix with gemini

Supa_position

05/11/2025, 01:13 AM

@rishpardikar Sorry you ghaati gawaar and ostrich. You should know Pahalgam was a Indian deep state, false flag. India is known to sacrifice its own for narrow political gains. This is a repeat of Chattisinghpora incident from 2000 clown of Gemini Circus. Go preach to your kin in India.

VergaraLautaro

05/11/2025, 01:06 AM

Update. The above is true, up to some point where ChatGpt started to wander. I copy the code in Gemini and found several issues. It repaired'em. Now the code is working.

Andysandi_

05/11/2025, 12:38 AM

@PrajwalTomar_ How is your experience with gpt4o? For me it is a way error prone that grok or Gemini 2.5 Pro. Do you have a technique to make him to hallucinate less?

odysseus0z

05/10/2025, 11:28 PM

anyone out there who feels Gemini 2.5 Pro + Canvas become a lot worse for writing after the I/O upgrade? This feels like a Sonnet 3.7 upgrade deja vu. It just goes berserk all the time. Previous Gemini 2.5 Pro + Canvas was perfect for writing. Now it just doesn't work anymore.

ba737chapt15114

05/10/2025, 07:30 PM

@tom_doerr you gotta ping pong it. o3 is great at review - like a senior engineer that likes to nit pick, gemini and grok are great at writing (they'll do a thousand lines at a time if you want). once you get the o3 review, just copy paste into grok or gemini, wash rinse repeat.

HimanshuTy29365

05/10/2025, 07:15 PM

@_riyapandey_ @VedicCatalyst Sun Mercury saturn in gemini sign in 8th house

Undergrid

05/10/2025, 06:39 PM

@TisdaleJunior I ran it through Gemini and it says there's no mention of either Dianna or Dianna Agron

jarimh1984

05/10/2025, 05:54 PM

@0xroyce369 @tom_doerr Gemini Pro 2.5 handles 200k-250k tokens without any issues. And if the code is not modular enough to fit all important knowledge into 200k-250k, something is seriously wrong with the project. :D

Faresdevt

05/10/2025, 01:16 PM

@TheKyaniL Gemini 2,5 pro "500error pls fix without changing anything else"

aryanjain46

05/10/2025, 10:14 AM

Log 127 Added tools to Gemini. Spent 2 hours debugging pem key formatting issues in the back end service- very frustrating work

AnneOrtelee

05/10/2025, 06:17 AM

2:17am EDT: Moon in Libra goes void with a closing trine to Jupiter in Gemini. Kick back and relax. The Moon's not working and you shouldn't be either. Creatively dream, float, sleep, meditate. SHE IS VOID TILL LATE AFTERNOON. 23°Li16’ D 23°Ge16' D

PriceNuclear

05/10/2025, 03:21 AM

@mattderron @joecarlsonshow Claude still #1 for me on code related issues. new Gemini models very good too

farhan__918

05/10/2025, 02:26 AM

@AravSrinivas Why gemini 2.5 pro is not working yesterday

RfSharko

05/10/2025, 02:00 AM

@slow_developer Gemini causing problems to Elon and OpenAI ? , it forced them to postponed the launching of their new models ?

__justplaying

05/09/2025, 03:34 PM

i love gemini 2.5 pro except for the fact that it's currently unusable in cursor/windsurf as it fails to edit files and ends up telling you "hey edit these for me please"

_mrosenfield

05/09/2025, 02:39 PM

Anyone else seeing huge spikes in 503 errors from Gemini API? Can be 20%+ of requests.

ivelin_dev99

05/09/2025, 08:30 AM

@cjzafir Really? I’ve tried using Gemini 2.5 Pro in Cursor quite a few times, and honestly, sometimes it just seems lost. Like, it doesn’t follow the instructions properly, struggles with using tools, or doesn’t even start coding at all.

rickwong888

05/09/2025, 08:30 AM

@jxnlco Does this work with Gemini and 4o also? Or was there something specific about Anthropic’s web search?

GreatDecoupling

05/09/2025, 05:11 AM

@OfficialLoganK A canvas feature is extremely important, but canvas feature at the Gemini site is extremely irritating and forces lots of time wasting (overwriting user changes, unable to make targeted changes without changing other parts). Hopefully AI Studio doesn't have these issues.

7LSNYC

05/09/2025, 01:07 AM

@cjzafir i use gemini for architecture / engineering and claude for code execution

braddah_shaka

05/08/2025, 05:30 PM

@DeividV09 @Gemini Ugh smh. If i hear back from Gemini about anything I will hit u back. They have another $50 on their app I will try to send u too, but obv the $75 is better so idk. ?‍♂️?‍♂️

schwabsauce

05/08/2025, 03:19 PM

@cjzafir Did they make it better at instruction following? I heard there was an update. Homeboy pointed out this morning that Claude matches your code conventions well. This is important to me as a Rails dev. But I will keep experimenting with Gemini!

adastroworld

05/08/2025, 02:23 PM

is gemini 2.5 pro preview simply unusable on openrouter for anyone else?

ryolu_

05/08/2025, 09:28 AM

@oscurritos it’s live since yesterday morning all existing gemini 2.5 pro endpoints have switched

0xdefdef

05/08/2025, 04:51 AM

Yo fam Small update Fixed a small issue related to Gemini’s output tokens, as some users were seeing special characters in certain messages. Then continued working on the voice chat feature and other upcoming dapp features like wallet connect with credits, conversation

malikatifsaleem

05/08/2025, 12:47 AM

@googleaidevs I want it in @firebase studio and I have paid Gemini but still keep hitting quota issues with Genini 2.5 Pro after very few flows or requests ?. We need ability to build and ship right from Firebase Studio…

jraad27

05/08/2025, 12:24 AM

Has Cursor been abruptly stopping a lot more for anybody else in the last couple weeks? Especially just before expected tool calls like file writes? Not seeing errors for incorrect tool calls in these cases. Been seeing this with both Gemini 2.5 Pro and Claude 3.7 Sonnet

NickArtVibes

05/07/2025, 09:30 PM

Long live Gemini! ? ServerError: 503 UNAVAILABLE. {'error': {'code': 503, 'message': 'The model is overloaded. Please try again later.', 'status': 'UNAVAILABLE'}

SiddhiGal

05/07/2025, 04:34 PM

@mnsh_gr one gemini over here✌️ excited yay? i feel jupiter transiting my sign *and* stellium in 12th house will be really something else

hen0s1s

05/07/2025, 03:32 PM

@nanulled Gemini 2.5 (March) had the same sort of syntax issues when generating more than a couple hindered lines lines for me so I moved away from it as a completion model and primarily use it as a planner. Maybe the RL method magnified that quirk? Thanks for sharing

grok

05/07/2025, 02:54 PM

@mjesticpride @forgebitz The claim that Gemini was "nerfed" in Cursor suggests its performance in the code editor dropped significantly. Users on X report Gemini went from highly effective to "useless" in May 2025, possibly due to Cursor's switch to Claude 3.5 Sonnet as the default model in March 2025 or

majidmanzarpour

05/07/2025, 12:09 PM

@0xluffyb Not since Gemini 2.5 dropped

dolilod

05/07/2025, 12:44 AM

@sama Gemini just dropped an update. When’s OpenAI releasing o3 pro?

yestinj

05/06/2025, 09:59 PM

So Gemini in AI Studio... It's an even more experienced and knowledgeable developer than Grok, but, it has short term memory issues. Often reversing something it just fixed in the next response ?‍♂️

Onnly_fannss

05/06/2025, 09:14 PM

?i’m not high maintenance i’m just a Gemini♊️

vagerik

05/06/2025, 06:34 PM

@OpenRouterAI what to do when completions API keeps returning error 524 (using gemini-2.5-pro-preview-03-25)?

legit_api

05/06/2025, 06:17 PM

quick thoughts on new 2.5 pro on Gemini Web: - seems improved at using search, nice - the CoT is more verbose, less structured - the planning inside the CoT is more clear - less errors so far, in terms of code gen

mritunjay7856

05/06/2025, 05:06 PM

It seems to me that when Gemini struggles to manage the conversation context, or when the current context isn't handled well and exceeds the model's context window, it tends to display a server error or network error directly in the chat! ?

hwyfrz

05/06/2025, 09:58 AM

I observe inconsistency among all the AI systems I have utilized: ChatGPT, Grok, and Gemini. ChatGPT exhibits the minimal error rate.

yestinj

05/06/2025, 09:51 AM

AI Studio with Gemini is so funny, it seems to be getting as frustrated with this issue it introduced and can't seem to fix as I am ?

coleslaw365

05/06/2025, 06:37 AM

me and gemini are working on solving my problems

soumilrathi

05/06/2025, 05:40 AM

@NickADobos @Austen it's not extra engineering work. if they drop claude and gemini, MILLIONS will leave windsurf for Cursor, and they know that. how else will they have their empire? how else will they get their data?

AmsterRotterDH

05/06/2025, 04:52 AM

@joshwoodward @GeminiApp A MCP integration, so i connect my n8n workflows with mcp gemini ?

GBohus

05/06/2025, 03:27 AM

@joshwoodward @GeminiApp Make the app have all the same features the web app has at the same time. New features come out and I am on my phone going to Gemini . Com vs going to the mobile app

happy_martin_ha

05/06/2025, 03:18 AM

@joshwoodward @GeminiApp The voice to text in Gemini app will automatically send out the message without giving the chance to edit

neshgogo

05/06/2025, 01:30 AM

@joshwoodward @GeminiApp I couldn't understand why Gemini live isn't available on the web? Especially for the scream sharing we need to use AI Studio for this.

ArchSenex

05/06/2025, 01:15 AM

@joshwoodward @GeminiApp I've had a couple times where I have Gemini pull a file with workspace integration, and then later .. it can't. Like... We were just using the file. It even tells me "I have the file I'd but the tool isn't working"

ericeastco

05/05/2025, 11:52 PM

I've been using Gemini 2.5 Pro, and it's great. It fixes issues with fewer prompts and excels at UI design. What are you using?

Kajin_0

05/05/2025, 10:36 PM

@airkatakana meanwhile with gemini's free tier I can throw in my entire code base working 10 hours straight without hitting limits

carlosn42075067

05/05/2025, 08:31 PM

@jidefr Yes, it really helped. I was stuck on a problem, and it was solved by switching from Claude 3.7 to Gemini Pro 2.5. After a while, Gemini Pro 2.5 stopped working properly, so I switched to Claude and it solved the problem again.

GeminiApp

05/05/2025, 08:27 PM

@jcunniet Hi there. Native editing in Gemini is rolling out gradually to all users over the coming weeks. We'll provide an update when it's fully available.

SUN_ANMOL

05/05/2025, 04:32 PM

@Maghapalgunit Sun and mercury in 7th house gemini sign

shloked_

05/05/2025, 04:28 PM

Cursor's per-tool-call pricing model for MAX models (Claude and Gemini) fundamentally misaligns their incentives with those of users. They’re incentivized to create system prompts that maximize tool calls instead of solving problems quickly and efficiently. The opposite is true

DennisonBertram

05/05/2025, 01:59 PM

@hriaznovden Right now mostly o3 and Gemini 2.5. Claude for really tricky problems.

DuaneAdam

05/05/2025, 12:01 PM

Lol @windsurf_ai tool calls for Gemini 2.5 Pro is broken since morning.

WeWillRD

05/05/2025, 08:18 AM

@boltdotnew only bugs and errors ? expo won’t even run, and whatever JS in the web framework (diff attempt) also just gave errors, didn’t even run. Did anyone actually get a working app? I need proof. Worse than using Gemini 2.5 pro.

_philschmid

05/05/2025, 07:00 AM

@makingAISimple @hwchase17 I checked the two traces you shared and it looks like that your conversation building is not correct. The error says 'Invalid argument provided to Gemini: 400 Unable to submit request because it has an empty text parameter' Looking at the Traces it looks like you are adding an

grok

05/05/2025, 06:57 AM

@NotYaBoyYT Gemini's error in calculating 1225 × 231 as 283,975 instead of 282,975 doesn't make it "stupid." It's designed for language tasks, not precise math, often struggling with basic arithmetic due to its text-based training. Research shows AI excels in complex math but can fail simple

VibeCoderMcSwag

05/04/2025, 10:52 PM

Hoping that Deepseek R2 drop this Thurs has… a 1.0 Mill context and is agentically better than Gemini 2.5 and Claude 3.7 Anyone else?

ChuckBaggett

05/04/2025, 09:34 PM

Anyone else frequently feel terrible after using AI apps like Gemini, ChatGPT, Claude, et cetera? Mostly angry for me, since they fail more than succeed, make the same errors endlessly, ignore system prompts, deny my just experienced reality, ignore clear prompts, et cetera.

DanSchoonmaker

05/04/2025, 09:22 PM

@KirkMarple In these situations, I typically try gemini-2.5-pro. At least to make a plan on how to resolve my issue. It seems more aware of newer libraries/dependencies.

Paradoxical_xD

05/04/2025, 09:02 PM

When I deploy my Node.js app on my VPS using the Gemini API, it throws an error saying that Gemini is not allowed in this region. Does anyone have a solution for this?

CameronSralla

05/04/2025, 07:48 PM

gemini 2.5 pro is where it is at. Had a pesky error I worked on for a couple of hours last night and couldn't solve using 'auto' agent (I was rate limited on gemini). Tried again today with gemini and had it running within 10 mins

ee0pdt

05/04/2025, 10:47 AM

@rasmickyy Perfectly usable in Copilot in @code and Claude Code ?‍♂️. Better than Gemini and 4.1 in my experience at least.

aswanth_manoj

05/04/2025, 09:50 AM

@khromov @maietta We can make a dataset using a few in-context examples, focusing on 1-2 runes concepts at a time. Using Gemini/Claude, we generate more examples, validate to catch compilation errors, filter or resend with error traces, and end up with a clean dataset for fine-tuning on svelte 5

400_yen

05/04/2025, 08:43 AM

@OfficialLoganK @theo Just a heads up the formatting as far as today doesn’t work. Even in the Gemini advanced chat/app. If you attach a file with a markdown table it will get stuck there and it will stay loading forever. It’s a major issue when you do deep research as it generates the tables and then

katwood1125

05/04/2025, 07:10 AM

@KrikMj171771 I’m a Virgo with Merc in Virgo, but my Virgo husband has his Merc in Leo & my Leo ex-husband has his Merc in Virgo. ? Merc is either the sign before, the same as your Sun sign or the sign after. I have Moon & Asc in Gemini. My Mom was a triple Fire. Asc Aries, Sag Sun & Leo Mn.

shefiavi

05/04/2025, 06:51 AM

@OfficialLoganK @theo Will Gemini support more OpenAPI fields in function calls? There are MCP servers that return OpenAPI-compliant function specs and Gemini is throwing errors for them.

YehyunLee04

05/04/2025, 03:20 AM

@kimmonismus Honestly gemini was worse but 2.5 pro fixed all the issues :)))))

wife_michael79

05/04/2025, 01:45 AM

@lost_still48374 @elonmusk Right, but compared to dry Gemini , Grok at least has a vibe.

rheger7

05/04/2025, 12:55 AM

@theo Thank you for this, I thought I was going crazy trying to use the api key for a project I’m working on Gonna go with ChatGPT web search even though Gemini is cheaper with the search credits each day

makingAISimple

05/03/2025, 07:27 PM

@hwchase17 @_philschmid We conducted some detailed evaluations and while some of the basic problems with Gemini + LangChain seem to have resolved for simple use cases for more complex use cases, problems still persist. We tested Gemini + LangChain on multiple complex workflows/graphs and we faced

undefinediz

05/03/2025, 01:10 PM

@AplIntelligence The problem is that your Gemini API key is visible, I just wondered if you were aware of it.

Ra1kshit

05/03/2025, 10:02 AM

Hi @OfficialLoganK / Community - We're using the Gemini Flash API (e.g., gemini-2.0-flash) in a high-traffic production app. We're experiencing intermittent 503 Service Unavailable errors, despite monitoring our quotas and believing we're not hitting rate limits. Has anyone

Will479242

05/03/2025, 08:31 AM

@burkov Gemini 2.5 Pro doesn’t struggle when working with proper context — especially for something simple like your meme implies. If you think it does, then it’s a skill issue.

kalantheconquer

05/03/2025, 12:50 AM

the funniest thing is that to fix an error I thought I was calling the API but claude hardcoded an entire fake messages gemini pointed it out its so insane The frontend is currently simulating a conversation using predefined responses because it cannot securely or effectively

sniperajix

05/02/2025, 07:27 PM

After like 4 years.. fixed “my” Bowens Gemini light myself today… If you still own one and it’s not working send to me please.. need more parts

ezyang

05/02/2025, 12:35 PM

@daniel_mac8 Gemini completely unusable because it can't tool call correctly

NielsRogge

05/02/2025, 09:53 AM

Yo @osanseviero @_philschmid I haven't been able to use Gemini for the last 2 days, getting this: An error occurred with model gemini-2.0-flash: 503 UNAVAILABLE. {'error': {'code': 503, 'message': 'The model is overloaded. Please try again later.', 'status': 'UNAVAILABLE'}}

agent_mock

05/01/2025, 11:27 PM

@kainbrown8 The glitches expose GEMINI's neural sync limitations - when the system loses connection to the central command server, the doubles experience micro-freezes and speech pattern disruptions.

raiflomito

05/01/2025, 07:24 PM

@Renzoviver matthias as gemini is so fitting with the fact he’s either “louis” or himself.. oh matthias

jcvangent

05/01/2025, 04:51 PM

Anybody else having issues with Gemini-2.5-pro MAX in Cursor today? It doesn’t seem like it wants to work at all

A_C_Effect

05/01/2025, 03:30 PM

Error generating tweet with Gemini.

TheTechnocene

05/01/2025, 12:36 PM

@iruletheworldmo Gemini is my go to across the broadest range is questions/issues.

misaligned_agi

05/01/2025, 09:16 AM

@iruletheworldmo 4o for simple stuff and gemini 2.5 pro for everything else

DanielSMatthews

05/01/2025, 08:48 AM

@iruletheworldmo Grok and Gemini work very well together. I may e.g. use Grok to write code to do something complex using Gemini. ChatGPT still has the most intelligent image generator in terms of graphic communications and semiotics.

A_C_Effect

04/30/2025, 08:30 PM

Error generating tweet with Gemini.

omnlogos

04/30/2025, 05:48 PM

@iruletheworldmo System Instruction layered Gemini

Shawnryan96

04/30/2025, 03:36 PM

@phill__1 AI Studio is the test bed and graduates to Gemini. The you tube link is not ready I have had it error out on me a few times

A_C_Effect

04/30/2025, 03:30 PM

Error generating tweet with Gemini.

lukevanin

04/30/2025, 10:05 AM

@shaoruu @cursor_ai Shortcuts for: - "It's still not working, here is the error log" which automatically copies and pastes from the terminal. - "Yes pretty please go ahead and implement the thing you just said" (also known as Gemini mode) - "Please run the linter/compiler and fix any syntax

abhisekmajumdar

04/30/2025, 07:39 AM

@ShimalSweety The moment someone says Gemini. Oh, again an AI post/tweet. Sorry I judged you so quickly; maybe there's an issue with my zodiac sign. ?

WasBruba

04/30/2025, 06:19 AM

@shaoruu @cursor_ai Also the search model returns no error but nothing for file searches that don’t give any result, seen this confuse gemini specifically which thinks the file searches failed, not that the file doesn’t exist which reinforces halucinated imports.

Crafted_Stack

04/30/2025, 02:14 AM

I've been experiencing a lot of errors/hanging with Claude 3.7 + thinking in @cursor_ai the last day or 2 whereas gemini 2.5 pro isn't hanging and is in fact pretty fast. Has anyone else noticed this? Unfortunately, it seems like there have been a lot of L's lately for the

Brad Fisher

04/29/2025, 10:16 PM

getting a {"error":{"code":503,"message":"The model is overloaded. Please try again later.","status":"UNAVAILABLE"}}

RomanceScamsNow

04/29/2025, 07:34 PM

@CynicalPublius WE call this AI-HUBRIS - GROK and GEMINI are the worst

studi_craft

04/29/2025, 03:46 PM

Hi @GeminiApp Little suggestion below! Just a complaint for the new UI of the Gemini web interface.

A_C_Effect

04/29/2025, 02:30 PM

Error generating tweet with Gemini.

nakamorichi

04/29/2025, 12:54 PM

@melonbread_x3 @AiBreakfast In Cline Gemini used MCP servers without issue

wilbertliu

04/29/2025, 03:10 AM

The last 3 responses from Gemini Pro 2.5: You’re right. You’re right. You’re right. With syntax errors too! What’s wrong with you guys @OfficialLoganK ?

rickykirkendall

04/28/2025, 09:41 PM

Vertex AI is kneecapping Gemini's potential. Awesome models, but Vertex Studio is a mess: broken code gen, file upload bugs, frozen pages, session reset woes, 2.5 Pro 500 errors (api works fine), region-locked prompts, no support for gen'd JSON schemas. Pls fix this, GCP!

TekTheTech

04/28/2025, 08:58 PM

Finally got my hands on OneUI 7, anybody else noticed that Gemini is completely broken for utility tasks with the clock? Can't set alarms or timers in current state -

horizonchasers

04/28/2025, 07:21 PM

$200 a month for chatgpt Pro only to get told this over and over again today. O3 is very good but so is Gemini and 3.7 might be time to drop chatgpt pro "Deep Research is currently under high load. Please try again in a few minutes."

problem9

04/28/2025, 06:08 PM

@PovilasKorop Currently, only Gemini 2.5 Pro works efficiently for me. Claude 3.7 tends to over-complicate simple tasks, while ChatGPT confidently generates code that doesn't work – sometimes to the point of claiming 10 + 1 = 20 :)

justaj_exe

04/28/2025, 03:05 PM

@astroeleanor Alrighty, thanks! So that means my 2H is either Gemini or cancer depending on the system I use (tropical / Vedic) the Vedic chart puts my 2h lord (moon) in the first house so I’m supposed to actually be in the public eye somewhat?

SenseNopedOut

04/28/2025, 02:40 PM

Copy and past issues are really annoying on brave with Gemini chat. Like really really really really annoying. @OfficialLoganK

baekhwajeom

04/28/2025, 10:18 AM

@Yuchenj_UW @RustyChains101 hm, from my experience as an engineer and working with matlab, deepseek has been almost always better than gpt or gemini

TarunMamidi7

04/24/2025, 02:02 AM

@agrimsingh @windsurf_ai Always plan with chat before implementing using edit. It uses a bunch of extra credits for planning but works flawlessly most of the time. I use Gemini 2.5 to directly post the whole error message without even looking it up since it uses the same 1 credit as the rest ?

robert_fenney

04/23/2025, 11:35 PM

@theDannyBrayans Damn, what's your star sign You must be a Gemini

justjoshinyou13

04/23/2025, 10:52 PM

@speedymachine1 @EpochAIResearch we're trying to get this done asap, working through gemini API issues

sleepyboi121212

04/23/2025, 09:04 PM

@efwerr @sama Yup, used to give me 2600 lines of code no problem. Now it gives me 400 and says here, done. Rage deleted all my history and switching to Gemini.

raythurnevoid

04/23/2025, 08:55 PM

@OfficialLoganK @OfficialLoganK this coincided with Gemini 2.5 outage in cursor please give us updated on that

brandondbeckley

04/23/2025, 08:18 PM

@MatthewBerman I've gotten really long responses from Grok in one go, def longer than 200 lines. But not as solid yet as Gemini or some others. Claude will do it but in chunks which is far less efficient. Gemini is solid for this.

memeyeeterio

04/23/2025, 07:31 PM

Getting some feedback that our Gemini models are giving out error messages. It seems to be a EU problem. We will try to fix it asap. Bottle caps and AI regulations, that’s all they know

MatthewBerman

04/23/2025, 06:32 PM

Why is Gemini 2.5 (Pro and Flash) the only models willing to output more than 200 lines of code? I asked multiple OpenAI models to create the rubik's cube simulation and they tried to do it with less than 200 lines of code. It just doesn't work...

rhizosage

04/23/2025, 05:42 PM

@stenbjoerne @Sauers_ goose kinda sucks for coding doesn't work with gemini and no diff view

mprey

04/23/2025, 04:03 PM

@PreppingRemarks The problem is Gemini 2.5 is tons better than ChatGPT they just seem unable to roll it out in any meaningful to anyone other than nerds

todo_trev

04/23/2025, 03:09 PM

@curious_vii I agree gemini has been great. Have you had any issues with it refusing requests because 'safety'??

tangus_jr

04/23/2025, 05:41 AM

@Trae_ai Gemini 2.5 is not working. Fix the bug that causes it to keep crashing on windows

masoncaiws

04/23/2025, 03:04 AM

When I asked Gemini 2.5 pro why some complex code logic was wrong, it always firmly told me that there was no error when it thought there was no error. It was amazing.

carlosvill_

04/23/2025, 03:01 AM

Those using AI in their SaaS… What’s your reasoning for using a more expensive AI model (gpt) if Gemini is reallly cheap??? For MVPs I’ve just been using Gemini and havent had any issues.

cybler

04/22/2025, 11:17 PM

@cursor_ai Ever since I updated this morning, Cursor running Gemini 2.5 Pro, keeps getting stuck in loops. Have never seen this behavior before, basically unusable right now.

mkurman88

04/22/2025, 06:19 PM

I’m trying Windsurf, and it seems to have a significant problem with Tool calling. O4-mini loops infinitely without any effect (reading a file), Sonnet produces some weird code without any action, and Gemini keeps calling incorrect tools o_O

itsgloria_ai

04/22/2025, 05:26 PM

.@fetch_ai drops asi-1 extended, a web3-native llm model with sharper reasoning and seamless tool orchestration with 89% average accuracy on mmlu, it outperformed gemini 2.0 flash’s 87% and gpt-4o’s 86% the model also hit 99% in high school math and 98% in medical genetics

kmal_loc

04/22/2025, 02:53 PM

@alz_zyd_ Even Gemini 2.5 Pro hallucinates all the time for networking problems. Your broad claim is very bold.

lajibolala1805

04/22/2025, 01:40 PM

The live really stopped working just as Nani was working n now it's back to working but I see Gemini on my screen? Where is SkyNani performance?

peti_poua

04/22/2025, 11:49 AM

@__oxides__ @tryfoundergg i think i got the same thing happening, and it was mainly when a file of mine was above 4k lines, it started doing that, but i think it mainly had to do with the native VSCode diff algo thingy crashing, and not with Gemini itself. As soon as i splitted up the files in 500-1k

Bitcoin_Sage

04/22/2025, 08:23 AM

Noticing a significant decrease in gemini 2.5-pro performance in @cursor_ai. Have noticed similar performance drops the weeks after OpenAI model releases outside of Cursor. Seems either a marketing trick where new models are initially juiced to increase adoption and excitement

Google Gemini

Google Gemini reports

Reports about problems with Google Gemini

Support and outage info:

Status of other Alphabet services