Google Gemini 2.0 is AMAZING - Realtime Stream TESTED | Multimodal API

[@r.m8146]

Amazing. Imagine having this kind of company to work and talk 24/7. It'll only improve going foward.

[@radektheplayer]

Only psychopaths do this, normal people think how to protect people's work

[@Jebble347]

Damn I can not deal with you never letting anyone finish a sentence. Even if it's AI.

[@alexanderrosulek159]

I am making an app for this rn using openai real-time it’s not hard anymore and ai voices about to be everywhere in products

[@TheTrainstation]

I tested it with cline and it fixed an issue I was having that sonnet and gpt 01 couldn't fix

[@AIInsights23]

Me too I try it with cline is real crazy time

[@loicbaconnier9150]

How do you plug it to cline please ?

[@kauasacramento2380]

@@loicbaconnier9150 get an OpenRouter key and select the model

[@pensiveintrovert4318]

​@@loicbaconnier9150I got an open router key, plugged that in, along with the url and model name. Sometimes it miscommunicated and one needs to press retry. It works BUT watch out for it overwriting files, have to watch it.

[@brto]

What was that, if you don't mind?

[@sidenote1459]

I think they had already been testing certain things in Gemini Pro, I've been using Gemini over Claude for most tasks just cuz it's gotten absurdly good in the last month seemingly out of nowhere. GPT and Claude still have their place, but Gemini is my go to for now

[@UselessDuckCompany]

I just tried using it to design a circuit and it did ok, although it seems to crash often. It was able to memorize the reference design and help me choose parts. It did not have perfect spatial understanding for the wiring.

[@mrd6869]

This is just scraping the surface. When this is smoothed out, i see i combining with Google Astra and have it take over your computer. 2025 is going to be very disruptive.

[@pauldolton9118]

my jaw is literally hitting the floor, this is amazing!

[@micbab-vg2mu]

Please test it more in agentic workflows :)

[@GlitterGuru]

Can’t wait to give it full mouse and keyboard control to just do the work for me 😂

[@ken5957]

There are so many models out now, for someone juat starting out whats the best approach to start with AI. Do u just pick one, whether it be Google, Anthropic, perplexity, openai etc and try to learn that really well? Or try learn a bit about all of them? If you were to pick one, which one would you recommend for people juat starting out?

[@DavidSegura99]

For now use Gemini 2 and google products going forward

[@zach4505]

Perhaps a comment line with special tag to pass as a prompt. Responses could also be an added comment. If that is too messy, then have a simple ui for prompts and responses like a google doc does for comments and editing.

[@luizpaes7307]

Yeah, it is really amazing. I am actually switching my gpt 4o endpoints for it

[@zeckliya7682]

The beginning of robot world conquest. 😂😂

[@minyakonga8897]

with ai studio's screen sharing, it can be a coworker, help you coding, that's amazing, like functionalities with cursor.

[@Ms.Robot.]

Smashing that code like a true AI code master.

[@matthewomalley4268]

Could this help evaluate existing website code to identify, for example, a list of JS variables within the script of the site?

[@justinwescott8125]

I think i would rather have text responses for coding. It's confusing listening to it try to tell you what to type. And why does it keep saying "backquote"?

[@gr8tbigtreehugger]

I have my own voice bot which also has the code base in context, so we can really talk about the code. I keep all the chat history in a terminal window in my IDE for reference plus copy/paste

[@grab_your_parachutes]

Cancelled my OpenAI sub today, they're going further off target under Sam.

[@DavidSegura99]

How can you request access to the other features?

[@CraftedLifeMoments]

pretty cool, love it :)

[@mrvarna-o5e]

hello i don't know why but i don't have this feature

[@wode9598]

It's good for a beginner. Don't know how much it can do.

[@grigrob9]

Hardcore drinking game idea: Drink each time he says "Pretty cool". Cheers!

[@aeramor]

hardercore: take a shot everytime the AI says `backquote`

[@zeckliya7682]

I love so much how he just simply cut the conversation 😂. Tell me again when the realistic robot woman type was installed with this type of A.i. Make sure there is a woman voice and the robotic version can be changed to a more realistic voice and can be dialect setting too. Might be time for me to remarry again.

[@rolestream]

The AI sounds slightly irritated half the time. O.o Otherwise amazing!! lol

[@EchoYoutube]

OpenAI must've decided to finally add Live Vision was because of the API release from Google. It came way too early. The Live Video is so poorly optimized like they never ran it through Docker. Older devices running ChatGPT just choke with the video function, it looks like Minecraft on a chromebook.

[@CmdrTigerKing]

its awesome but it needs a better way to communicate tasks/steps....

[@Julian-tf8nj]

can one switch the voice? For example, to a female voice?

[@Darkurge666]

This doesn't really feel efficient compared to just typing. I like that you can talk to it, but there must be more useful use cases where it is a better fit. Quoting code and typing isn't exactly efficient. :D

[@AdvantestInc]

This breakdown of Gemini’s real-time capabilities highlights its potential for streamlining workflows. It’s exciting to imagine how this tech could reshape development pipelines.

[@melbar]

Soloppgang is a funny name for a city ;-)

[@agatitytube]

It looks truly amazing, though it's very slow. But I can imagine this in a year or two...

[@johnne86sd]

I would say please slow down when you're are telling me what to type and when you need to explain something talk at normal speed. I wonder if can so that.

[@timber8403]

Speech not even close to Chatgpt

[@phdcosta]

wow! Could you help a 3rd world programmer to choose which platforms to subscribe? our currency is 6 by 1 dollar, so price is a challenge for us, every investment is 6x expensive. I'm a teacher, with PhD, salary 1200 dollars month. As you can see, cost of AI platforms is critical for us... free solutions are far from OpenAI and Google

[@thenoblerot]

Google AI studio offers a lot of free credits

[@iuliusRO82]

Could you please mute "Hey Google" phrase next time? You activated my nest mini and my phone 😂. Thanks

[@LibertyRecordsFree]

Honnestly, i am not impressed... It is not a real product, still experimental, and in term of coding, it does not do better. I tested for complex task and it is behing gpt-o

[@johntor-yr3gh]

very poor

[@MrErick1160]

I tested it it was actually horrible in my opinion 😅. LOTS of bugs in it's language and speach is quite annoying

[@PseudoProphet]

Cope harder and pay another $200 dollars to OAI next month as well. 🤡😂😂

[@angrybeast6387]

It's still in experimental phase just wait for next month to get it full released with pro model

[@angrybeast6387]

​@@PseudoProphet😂 yea ai studio is much better and all free

[@MrErick1160]

@@PseudoProphet not paying shit bro, not coping for shit as well, in the state AI is right now, those agents are useless in practice, wether it be google OpenAI Microsoft or Claude. Agents are not ready for smooth conversations or complex multi step tasks. Sorry to crush your AI-dream 😅🥲