Perfect, this is the type of videos i want to watch every week, latest and greatest new generative AI models. Learned about Hunyuan 2, great find, thanks
Your best news for AI i know of. I don't even know how you find all this stuff. I kind of just live on Comfyui these days so if it's not on comfyui i don't often hear about these. the first things in this video for example i'll likely wait until comfyui support to try it, i am not sure i feel like figuring out how to install more standalone things with often a bad UI or command prompts. But your AI News is so well done, really.
@@jonasvq comfyui is an AI UI that let's you use a lot of various AI tools in one really amazing and powerful UI. Like Automatic1111 or whatever but different. I think comfyui is way more powerful, efficient then all rest out there.
VideoFX is off the darn scale, I can actually use this to create almost any picture I require to use in my videos, This is Ai gone crazy. Thanks for the tip.
@@theAIsearch wow, the fact you are hoping it slows down shows just how fast everything moves in this space. Respect the grind, but chill wait at least until we get the neuralink for faster video editing.
I’m ironically relieved at the amount of different models and especially opensource models there are popping up. The fear was that AI would get monopolized, dooming us to serfdom, but the fact that it will likely be readily available to everyone actually just has the potential to increase everyone’s capacity…. Hopefully anyway
Yes it's a tool and should only be used as a tool. People always love making the excuse of worse case scenario to get rid of it. It's a terrible argument. There will always bad actors, blame the bad people not the tool. It's common sense. The positives seem to outweighs the negatives so far, I see that as a win in my book. Just imagine the breakthroughs for curing diseases! Although I could be totally wrong and AI just wipes out all of humanity due to the rule of equivalent exchange.
if something is given to the public for free, there isnt going to be any profit made directly from it; you can hope to generate tons of new jobs from this open insemination, but the problem is that it is still to be demonstrated these jobs cannot be automated themselves; but even if there is a boom economy, the work time needed to produce the services with these ai tools will drastically contract, with profit collapse and an explosion of goods on the market, and there is the serious risk of saturation and impossibility to absorb them. And overproduction and deflation, historically, leads quickly to protectionism, commercial wars and finally military conflicts to destroy fixed and variable capital and then redistribute the profit quota. This means not only serfdom, but wipeout of mankind. We already see these symptoms today, when ai and robots are at the beginning; look at the automotive industry, where mature western companies are virtually already failed due to extensive production automation. Robots and machines per se, without human exploitation, cannot make any profit at all in the average cycle of production and distribution.
Timestamps (Powered by Merlin AI) 00:07 - New AI tools enhance video and 3D model generation capabilities. 02:32 - New AI tool generates detailed 3D models from images or text prompts. 07:08 - AI enables dynamic video generation and first frame editing with 3D and image prompts. 09:23 - Dream Catalyst enables creative editing of 3D Nerf videos through simple prompts. 13:55 - AI model demonstrates impressive object positioning and realistic visuals in generated images. 16:04 - New AI models offer improved image generation and video editing capabilities. 20:22 - New AI agent automates tasks across desktop and web browsers. 22:25 - AI agents enhance productivity in desktop applications and web browsing tasks. 26:46 - New open-source AI model allows extensive customization and use for commercial purposes. 28:54 - The AI operator requires user confirmation for actions, making it cumbersome. 32:58 - User limitations and functionalities of the new AI model are highlighted. 35:06 - AI generates precise images by combining multiple input features seamlessly. 39:02 - New depth estimation tool significantly improves accuracy in chaotic video scenes. 41:05 - Introduction of DeepSeek R1, a new AI model surpassing OpenAI's capabilities. 45:16 - 2025 promises exciting advancements in AI tools and news.
Really nice summary of the weeks events and cool tech. No hype, no speculation, no drama, like other AI creators. Just the news with tangible demonstrations of their capabilities. Thank you, I've subscribed, keep it up.
If There Was Only One TH-cam Video That Could Watch In A Week Then It Would Be Yours. The Level Of Excitement That I Get When I See A New Video Of You Is Unmatched.
39:23 This is the most importan news in AI probably in the last 2 years. Finally we could breakthrough the psychological barrier of 10 seconds AI videos. A promise that Sora couldn’t deliver. For people who don’t know it, depth maps are the core of AI video models.
Absolutely do a full tutorial on Ui-Tars, deep dive, a few different scenarios for setup, Windows, Linux etc. If I've seen anything it's locally hosted, ai controlling a PC is something thousands of people have an interest in. Biggest hurtle i see is vllm setup for people.
The hype in AI can sometimes feel like the emperor’s new clothes-buzzwords, overblown promises, or exaggerated capabilities can create an illusion of groundbreaking advancements, even when the actual substance may not measure up. However, just as in the story, there are often those who point out the gaps or limitations. The challenge lies in distinguishing genuine innovation from empty hype, as the true value of AI lies in its practical, impactful applications-not just the spectacle. When billions are being pumped in, the pressure to deliver, or at least appear to deliver, can lead to a cycle of overpromises. Investors chase the “next big thing,” and startups or companies might exaggerate capabilities to secure funding. This creates an ecosystem where hype sometimes overshadows real progress, and the “new clothes” are sold before they even exist. It’s not to say there isn’t meaningful work being done, but the gap between marketing and reality can grow wider when so much money is at stake.
insanely good= proceeds to show something that we had for half a year now, WOW such a big improvement. I am 3D artist i dont see anything taking my jobs 5-10 years :,D
Wow Dream Catalyst didn't die. They showed that off about a year and a half ago, having made insane progress at the time they showed it back then compared to any other tech (imo) but then went radio silent. Glad to see it manifest into actual released code.
At first I was like, dude, I don't have time to watch a 45 min video on just AI news. After the video is finished I am wearing my surprised pikachu face wondering where to start testing all this new stuff. I'll just figure out the sleeping part in a few days.
That first Frame editing I'd SUPER innovative. I was literally was just talkin about how it would be cool to edit a Frame through photoshop & have the original video act as if that edit you made was always there. Next stage is making it work with changing people/faces in a frame (if u can't already lol)
Really enjoy these videos, the only downside is which one do I build out first! I sure hope UI-TARS releases a linux desktop version as it only supports Win/OSX
Fascinating and thanks! It would be very informative if you can make a video explaining the different AI training method : RL, supervised Training, Value Functions, Chains of thoughts, Monte Carlo Tree Search and how these lead to real reasoning.
In the OpenAI Operator AI agent segment, you asked why Operator used Bing instead of Google. Microsoft is a major OpenAI partner, they're leveraging the brand synergy.
SVECTOR is tech company from India, creating advanced AI Models and Incredible ai they are really passionate about doing this and also there reasoning model is open source..
18:00 is really not a flaw at all. We can see the reflection was not selected for deletion so the tool worked as expected by leaving it untouched. If it had deleted the reflection I would have considered it a flaw, cause it would be deleting more than I asked it to. Imagine I wanted to make a scary movie where something can be seen on reflections but not directly, for example. If it always deleted reflections I would not be able to achieve such effect.
image all that progress didn''t take place is just the last year but would have streched over 5 years over even 10. It's like going from the NES to PS4 in one year!
AI visuals may be the best use of AI for peaceful means. Humans become art editors for quality control. Artists need tools that can give story tellers choices.
This video shows how AI can now create 3D models, change textures, and more with just a simple prompt and a few clicks. People who have put in years of effort, developing plugins for 3D software like Blender to improve texture placement or other tasks, are already seeing their work rendered obsolete by AI. Now, anyone can achieve what they’ve worked so hard for with little effort. In the short term, even the most advanced professionals are at risk of being replaced. While some may be seduced by the idea of enhancing their performance with AI, they fail to see that, very soon, they’ll be completely erased from the map. This rapid evolution will lead to the loss of many jobs, skills, and meaningful work. In this context, I find it hard to see how AI could become our friends in the near future. For example, when an OpenAI executive announces, in collaboration with Trump, that they are going to conduct research to produce cancer vaccines in a matter of seconds, I can't help but think we're in deep trouble. Instead of investigating the origins of these turbo-cancers, which exploded so suddenly right after the numerous injections in 2021 (what a coincidence…), they’re already telling us that new injections will be made to cure the effects of those initial injections. I don’t know if I’m talking to members of the Matrix, some kind of Agent Smiths, but it’s so obvious, yet…
Imagen is really amazing. Ive been using it in my workflows since ver 2. I wish that google would release veo 2 but I think they still need to generate buzz
Pls do a video on UI-TARS and how to download it. it's funny how china AI model are beating some of the best AI models out there and china are making most of there AI for free....nah they are the goat bro.🐐🔥 And thanks for the videos I want to try some of these AI keep up the good work.
Don you have the feeling that OpenAI is releasing things that somehow look old? Already seen somewhere else? Is this the visionary company that will make the US the world capital of AI? Running always from behind?
First things first, I really liked your video! However, I wanted to point out that the rankings you’re showing don’t seem entirely realistic. The number of votes is always lower for the winner, which suggests less precision in the results
Using the google model, try making an elephant balance on a man's head. or, any image using a superhero. Or a grand piano balancing on top of an umbrella.
When AI Video generations started I commented on that "in the future" we would ask the AI 'Make me an 80s style action movie, the good guys are Kurt Russell and Val Kilmer, the bad guy is Alan Rickman". It seems that this "in the future" is much closer than I thought. 😮 EDIT: Can you imagine? "AI make me a version of DIE HARD but the MC is now Jackie Chan." the future looks so exciting!
That DiffuEraser is exactly what I am after for my client whos crazy Karen X is trying to take him to court forhalf his empire after he caught her having an affair. With that software, we go pow and presto, and then what? Say what now? No your honor, we never have been in a relationship. Footage? Not a drop, Sir.
@theAIsearch I know right! Also, we could make a realtime missing moon app that takes your camera feed and vision models to recognise the moon when camera is looking at it, the moment it comes into frame the DiffuEraser makes it get gone in real time... Hilarious to show theorists in live app that the moon their eyes can see actually disappears in front of a lens
Can you show how to use UI-Tars for Desktop? I have installed it, and you can only select VLLM and it looks like its working but no matter what you ask it to do, it takes the first screenshot and then it stops. I cannot find any tutorial on how to set it up properly.
it's because your using a not visual llm I think. I have the same problem but have only huggingface choice or VLLM but no free API to use... I'am stuck !
thanks for useful information. Can you share some info where I could find Ubuntu version of UI-TARS ? Or, maybe you used ollama to run it, I just thought. I'll try ...
Perfect, this is the type of videos i want to watch every week, latest and greatest new generative AI models. Learned about Hunyuan 2, great find, thanks
You're welcome!
Your best news for AI i know of. I don't even know how you find all this stuff. I kind of just live on Comfyui these days so if it's not on comfyui i don't often hear about these. the first things in this video for example i'll likely wait until comfyui support to try it, i am not sure i feel like figuring out how to install more standalone things with often a bad UI or command prompts. But your AI News is so well done, really.
Thanks!
whats comfyui? and what would it do when it supports comfyui?
@@jonasvq comfyui is an AI UI that let's you use a lot of various AI tools in one really amazing and powerful UI. Like Automatic1111 or whatever but different. I think comfyui is way more powerful, efficient then all rest out there.
We eating good woth all these videos coming out, thanks ❤
You’re welcome!
Now next step is Agents..., how amazing is this that we are seeing this progress from start
VideoFX is off the darn scale, I can actually use this to create almost any picture I require to use in my videos, This is Ai gone crazy. Thanks for the tip.
3 videos in a row? I know AI never sleeps but I hope you're sleeping well, then you end up sleeping on the toilet
lol thanks. hope it slows down so i can get some rest!
@@theAIsearch wow, the fact you are hoping it slows down shows just how fast everything moves in this space. Respect the grind, but chill wait at least until we get the neuralink for faster video editing.
@@theAIsearch 500 billion dollars tells me that it whohnt ;)
AI search sleeps on a pile of burnt out GFX cards
Looks like we're getting ever closer to singularity. Every day brings new stuff, and it's almost impossible to keep up.
yes!
@@theAIsearchsounds scary to me.
Not sure what singularity is, but I've never kept up, just use what you want. Don't think many use everything.
I’m ironically relieved at the amount of different models and especially opensource models there are popping up. The fear was that AI would get monopolized, dooming us to serfdom, but the fact that it will likely be readily available to everyone actually just has the potential to increase everyone’s capacity…. Hopefully anyway
Yes it's a tool and should only be used as a tool. People always love making the excuse of worse case scenario to get rid of it. It's a terrible argument. There will always bad actors, blame the bad people not the tool. It's common sense. The positives seem to outweighs the negatives so far, I see that as a win in my book. Just imagine the breakthroughs for curing diseases! Although I could be totally wrong and AI just wipes out all of humanity due to the rule of equivalent exchange.
if something is given to the public for free, there isnt going to be any profit made directly from it; you can hope to generate tons of new jobs from this open insemination, but the problem is that it is still to be demonstrated these jobs cannot be automated themselves; but even if there is a boom economy, the work time needed to produce the services with these ai tools will drastically contract, with profit collapse and an explosion of goods on the market, and there is the serious risk of saturation and impossibility to absorb them. And overproduction and deflation, historically, leads quickly to protectionism, commercial wars and finally military conflicts to destroy fixed and variable capital and then redistribute the profit quota. This means not only serfdom, but wipeout of mankind. We already see these symptoms today, when ai and robots are at the beginning; look at the automotive industry, where mature western companies are virtually already failed due to extensive production automation. Robots and machines per se, without human exploitation, cannot make any profit at all in the average cycle of production and distribution.
Timestamps (Powered by Merlin AI)
00:07 - New AI tools enhance video and 3D model generation capabilities.
02:32 - New AI tool generates detailed 3D models from images or text prompts.
07:08 - AI enables dynamic video generation and first frame editing with 3D and image prompts.
09:23 - Dream Catalyst enables creative editing of 3D Nerf videos through simple prompts.
13:55 - AI model demonstrates impressive object positioning and realistic visuals in generated images.
16:04 - New AI models offer improved image generation and video editing capabilities.
20:22 - New AI agent automates tasks across desktop and web browsers.
22:25 - AI agents enhance productivity in desktop applications and web browsing tasks.
26:46 - New open-source AI model allows extensive customization and use for commercial purposes.
28:54 - The AI operator requires user confirmation for actions, making it cumbersome.
32:58 - User limitations and functionalities of the new AI model are highlighted.
35:06 - AI generates precise images by combining multiple input features seamlessly.
39:02 - New depth estimation tool significantly improves accuracy in chaotic video scenes.
41:05 - Introduction of DeepSeek R1, a new AI model surpassing OpenAI's capabilities.
45:16 - 2025 promises exciting advancements in AI tools and news.
Synopsis: end of the humanity coming soon...
Really nice summary of the weeks events and cool tech. No hype, no speculation, no drama, like other AI creators. Just the news with tangible demonstrations of their capabilities. Thank you, I've subscribed, keep it up.
Thanks for the sub!
If There Was Only One TH-cam Video That Could Watch In A Week Then It Would Be Yours. The Level Of Excitement That I Get When I See A New Video Of You Is Unmatched.
Same. And he never has to tell us how shocked the industry is 🙂
Thanks!
39:23 This is the most importan news in AI probably in the last 2 years. Finally we could breakthrough the psychological barrier of 10 seconds AI videos. A promise that Sora couldn’t deliver. For people who don’t know it, depth maps are the core of AI video models.
Slow down before I report your channel for overwhelming braingasms! I am still reeling over how amazing DeepSeek R1 is! 😮
lol this week was crazy. hope it slows down a bit
Appreciate the no-nonsense way you present things. None of that drama and extra hype I see on other channels. I'm unsubscribing from those right now!
I'm not a bot
Absolutely do a full tutorial on Ui-Tars, deep dive, a few different scenarios for setup, Windows, Linux etc. If I've seen anything it's locally hosted, ai controlling a PC is something thousands of people have an interest in. Biggest hurtle i see is vllm setup for people.
Damn this year started very promising
yes!
The hype in AI can sometimes feel like the emperor’s new clothes-buzzwords, overblown promises, or exaggerated capabilities can create an illusion of groundbreaking advancements, even when the actual substance may not measure up. However, just as in the story, there are often those who point out the gaps or limitations. The challenge lies in distinguishing genuine innovation from empty hype, as the true value of AI lies in its practical, impactful applications-not just the spectacle. When billions are being pumped in, the pressure to deliver, or at least appear to deliver, can lead to a cycle of overpromises. Investors chase the “next big thing,” and startups or companies might exaggerate capabilities to secure funding. This creates an ecosystem where hype sometimes overshadows real progress, and the “new clothes” are sold before they even exist. It’s not to say there isn’t meaningful work being done, but the gap between marketing and reality can grow wider when so much money is at stake.
Thank you very much AI search . I really appreciate the channel.
Welcome!
Bro I just finished your previous video😂and this came out, nice 👍
Thanks!
God damn. This year has started with a bang
yes!
Outstanding research and delivery of information, 10/10
Thanks!
insanely good= proceeds to show something that we had for half a year now, WOW such a big improvement. I am 3D artist i dont see anything taking my jobs 5-10 years :,D
Wow Dream Catalyst didn't die. They showed that off about a year and a half ago, having made insane progress at the time they showed it back then compared to any other tech (imo) but then went radio silent. Glad to see it manifest into actual released code.
At first I was like, dude, I don't have time to watch a 45 min video on just AI news. After the video is finished I am wearing my surprised pikachu face wondering where to start testing all this new stuff. I'll just figure out the sleeping part in a few days.
Man you are amazing please keep up the good work! Love everything about your videos!
Thanks!
That first Frame editing I'd SUPER innovative. I was literally was just talkin about how it would be cool to edit a Frame through photoshop & have the original video act as if that edit you made was always there. Next stage is making it work with changing people/faces in a frame (if u can't already lol)
0:44 "H'yunyuen 3D two" 🗣🗣🔥🔥
I need to make more money for buy gpu because I want absolutely use everything from this video😅so very insane video and news report! Thanks
Really enjoy these videos, the only downside is which one do I build out first! I sure hope UI-TARS releases a linux desktop version as it only supports Win/OSX
Very stoked on your AI news cap'
Thanks
Это полный 42, столько всего невероятного крутого в одном видео, прям те нейросети о которых я думала по очереди все
The pace of progress is just insane.
3:12 "This indeed looks like a lovely rabbit" yeah.. rabbit with squirrel tail. Lol
This AI agent will be especially useful for people with disabilities if they add options with a fast voice assistant.
Fascinating and thanks! It would be very informative if you can make a video explaining the different AI training method : RL, supervised Training, Value Functions, Chains of thoughts, Monte Carlo Tree Search and how these lead to real reasoning.
In the OpenAI Operator AI agent segment, you asked why Operator used Bing instead of Google. Microsoft is a major OpenAI partner, they're leveraging the brand synergy.
Pro tip is use omnimatte to find shadows and other indirect scene interactions so diffueraser can erase even more seamlessly
Thanks Bro. Good coverage. Much appreciated
Thanks!
Yes please🙏🏾🙏🏾 make a tutorial on how to install UI tars. Online and offline
yes we need this
SVECTOR is tech company from India, creating advanced AI Models and Incredible ai they are really passionate about doing this and also there reasoning model is open source..
Great video as always ❤
Thanks!
Hunyuan3D-2 is now 3rd place on the 3d arena. both trellis and trellis 3dgs before
lol that was fast!
18:00 is really not a flaw at all.
We can see the reflection was not selected for deletion so the tool worked as expected by leaving it untouched.
If it had deleted the reflection I would have considered it a flaw, cause it would be deleting more than I asked it to.
Imagine I wanted to make a scary movie where something can be seen on reflections but not directly, for example. If it always deleted reflections I would not be able to achieve such effect.
Another INSANE video.
Of course.
You know, when everyone uses that word ALL THE TIME it really dilutes the meaning
Can you believe THAT!!!
are you SHOCKED or STUNNED yet
@@theAIsearch I'm totally desensitized is what I am. Also tired and depressed.
It's some sort of marketing, to open a repo on github, even though the model is not open sourced... I'm about that Kimi model... 😅😆
Thanks for your videos!
Please a tutorial how to install UI TARS AI Agent :)
Hi
Thanks for the amazing info, kindly share the steps for installing the UI Tars AI agent.
All the best
image all that progress didn''t take place is just the last year but would have streched over 5 years over even 10. It's like going from the NES to PS4 in one year!
AIs that can write email?
Get ready for an inundation of spam like you never could have imagined!
18:20 I can finally erase all the annoying characters in movies thx man
Please mac a video setting upi UI Tars on Mac. Thank you!
Great video
AI visuals may be the best use of AI for peaceful means. Humans become art editors for quality control. Artists need tools that can give story tellers choices.
This video shows how AI can now create 3D models, change textures, and more with just a simple prompt and a few clicks. People who have put in years of effort, developing plugins for 3D software like Blender to improve texture placement or other tasks, are already seeing their work rendered obsolete by AI. Now, anyone can achieve what they’ve worked so hard for with little effort. In the short term, even the most advanced professionals are at risk of being replaced. While some may be seduced by the idea of enhancing their performance with AI, they fail to see that, very soon, they’ll be completely erased from the map. This rapid evolution will lead to the loss of many jobs, skills, and meaningful work.
In this context, I find it hard to see how AI could become our friends in the near future. For example, when an OpenAI executive announces, in collaboration with Trump, that they are going to conduct research to produce cancer vaccines in a matter of seconds, I can't help but think we're in deep trouble.
Instead of investigating the origins of these turbo-cancers, which exploded so suddenly right after the numerous injections in 2021 (what a coincidence…), they’re already telling us that new injections will be made to cure the effects of those initial injections.
I don’t know if I’m talking to members of the Matrix, some kind of Agent Smiths, but it’s so obvious, yet…
Yes please do a tutorial for UI TARS desktop with a local LLM
Imagen is really amazing. Ive been using it in my workflows since ver 2. I wish that google would release veo 2 but I think they still need to generate buzz
nice! me too - i can't wait to get my hands on veo
Pls do a video on UI-TARS and how to download it. it's funny how china AI model are beating some of the best AI models out there and china are making most of there AI for free....nah they are the goat bro.🐐🔥 And thanks for the videos I want to try some of these AI keep up the good work.
ai never sleep heheh good intro you should to make it as song
Don you have the feeling that OpenAI is releasing things that somehow look old? Already seen somewhere else? Is this the visionary company that will make the US the world capital of AI? Running always from behind?
Many thanks man, ur the best
You're welcome!
Don’t sleep on Forbidden Gateway to Money Mastery! It’s one of those rare books that changes how you think forever
First things first, I really liked your video! However, I wanted to point out that the rankings you’re showing don’t seem entirely realistic. The number of votes is always lower for the winner, which suggests less precision in the results
Thanks for sharing!
Can't wait to see if UI-Star can install Github repose or make them into 1-click installs automatically
I'll need to read that UI-TARS paper. I'm curious to know how they did that.
What use cases of Video Depth Anything?
are there already some useful tools for generating animated avatars, cards, sprites in 2D?
A.I. in 2025 = Very advanced technology
To install the technology = use tech from the 1970's
Really good video.
Thanks!
Using the google model, try making an elephant balance on a man's head. or, any image using a superhero. Or a grand piano balancing on top of an umbrella.
8:21 not so smart - no book reflection on laptop screen :)
Thank you
You're welcome
When AI Video generations started I commented on that "in the future" we would ask the AI 'Make me an 80s style action movie, the good guys are Kurt Russell and Val Kilmer, the bad guy is Alan Rickman". It seems that this "in the future" is much closer than I thought. 😮
EDIT: Can you imagine? "AI make me a version of DIE HARD but the MC is now Jackie Chan." the future looks so exciting!
I just uploaded my original character on hunyuan and it's actually f-king great (better than Microsoft trellis so far)
nice!
That DiffuEraser is exactly what I am after for my client whos crazy Karen X is trying to take him to court forhalf his empire after he caught her having an affair.
With that software, we go pow and presto, and then what? Say what now? No your honor, we never have been in a relationship. Footage? Not a drop, Sir.
great idea!
@theAIsearch I know right!
Also, we could make a realtime missing moon app that takes your camera feed and vision models to recognise the moon when camera is looking at it, the moment it comes into frame the DiffuEraser makes it get gone in real time...
Hilarious to show theorists in live app that the moon their eyes can see actually disappears in front of a lens
Amazing magic....but I know it is not. Creation of conceptual reality programmed in natural language app or making selection.
FYI, I did the Komodo dragon test on Imagen, and it actually does get it right. So it must be in their training data.
thx for sharing!
china is leading in AI
25:19 Shows Elon Musk completely jacked in a superman suit
"And look how realistic this is!"
I know! Not very!
this guy is all i need
24:00 and if someone get a way to prompt remotely from outside it will be the end then ?
Did you notice how Google's image gen automatically mixed ages/races of the generated hands? 😂
Wokie's gonna woke.
This is the best channel for ai news
Thanks!
👋 hi 👋
hi
Can you show how to use UI-Tars for Desktop? I have installed it, and you can only select VLLM and it looks like its working but no matter what you ask it to do, it takes the first screenshot and then it stops. I cannot find any tutorial on how to set it up properly.
it's because your using a not visual llm I think. I have the same problem but have only huggingface choice or VLLM but no free API to use... I'am stuck !
A short tutorial on installing UI-TARS please.
❤ deepseek r1
thanks for useful information. Can you share some info where I could find Ubuntu version of UI-TARS ? Or, maybe you used ollama to run it, I just thought. I'll try ...
18:15 prompt: "erase all people"
AI response: "sure thing, commencing global thermonuclear war"
Is it just me or does this guy sound totally like the dude from the "What I've Learned" channel often talking about diets?
Please make a tutorial to install UItar
tokenverse looks brutal
make a detailed video on browser-use, how to install and use please
Let's gooooooo 🗣️🗣️
😃😃😃
G with the flow is cursed
Please do a UI TARS video!! 🙏🙏
China🇨🇳🇨🇳 and Brazil 🇧🇷🇧🇷 will dominate the world with the artificial intelligence sector.🎉❤🎉❤🎉❤
Hunyuan3d-2 only gives me terrible results compared to Trellis! Not got 1 good gen so far for any category. What am I doing wrong? :-(
Nevermind your example gens here are just as bad. Trellis is better (you have to edit the source to allow fine detail gen though).
nex step from China ist free KI Agents. that became the finale punsh to Open Ai.
Please do the full tutorial on UI Tars. It didn't work when I tried doing it myself
what impact will deepseek r1 have on ai voice changers? . are companys already out there use those??
Can we just take a little break
I want to too lol