Loading video...

Video Failed to Load

There was a problem loading this video. This could be due to a temporary network issue or the video might be unavailable.

Gemini 4 Pro has surprised again - A mechanical Bee :) It has improved a lot since the last checkpoint in terms of design taste and prompt adherence.. I have a few more outputs that I'll be sharing soon.. I'd been trying endlessly since the new checkpoint dropped, and... show more

Bee

4,149 subscribers

110,051 views • 7 days ago •via X (Twitter)

Science & Technology

Anya Rossi• Live Now

Private livecam show

57 Comments

Harshith7 days ago

maybe they pulled it i didnt get anything today

Bee7 days ago

Yupp me too since morning

Karthike7 days ago

This is so freaking awesome 💕 May we see a head-to-head comparison with Opus 5.5 and GPT 6 Astra, please 🥺

Bee7 days ago

Yupp alrdy in works

BitSharC7 days ago

Is it you?

Bee7 days ago

👀

Marcel7 days ago

Bee generated a mechanical Bee :) we really can’t wait anymore for 4 pro tho. taking too long 😭

Bee7 days ago

Mid oct.. im pretty sure

Mr. Oblivion7 days ago

@marcthecreatorr Tbh, I think depending on how fast it improves internally or how fast they get the safeguards right, they might release it this week. They released skills today too. Honestly its either this week or next week at the latest

Bee7 days ago

@marcthecreatorr Skills have been out for a week.. they announced it today.. But yeah they released the new checkpoint just 2 days after the last one so they're definitely moving fast

Ming Ming7 days ago

I'm really blown away by your results this time - it's absolutely insane. GPT can't even come close to it. I'm curious if your opus 5.5 results can beat it. Thanks for putting in all that time

Bee7 days ago

I'm gonna post the comparison soon..

Ming Ming7 days ago

I wonder how many minutes it took different models to run this task?

Bee7 days ago

Gemini did it in 21 minutes.. the rest of the timings I'll post it with the comparison but they all took much more time than Gemini

Harshith7 days ago

this output is dope 🐝🐝

Bee7 days ago

Bees buzzing

Ming Ming7 days ago

$GOOG I suspect tonight's stock price movement might be somewhat related to your results😂The result is exactly what I expected from Fable 5.5

Bee7 days ago

Hahaha lol

Doe7 days ago

Wowwww this output is actually kinda insane 😭

Bee7 days ago

I'll be comparing it with some other models soon!!

Doe7 days ago

I think it might beat Astra 👀

Bee7 days ago

Yupp ig so.. Will compare with it too

Doe7 days ago

can't wait

AshutoshShrivastava7 days ago

Google is coming hot...

Bee7 days ago

Hoping they don't nerf it..

Genysis AI7 days ago

Prompt?

Bee7 days ago

Will share it soon

ThisMightWork7 days ago

The warm paper and little specimen labels really make this. Feels like opening an old natural history book and finding out the illustration escaped the page.

its_okarun7 days ago

Damn finally google making comeback

@forHonor7 days ago

Holllyyyy beee|!

vinn17 days ago

its not pulled

Bee7 days ago

Are you getting outputs?

vinn17 days ago

yep, it gives an error like this for 3D or SVG tasks but works for creative writing or other low effort stuff

Bee7 days ago

Could uh ping me once if it starts giving outputs..

vinn17 days ago

sure

SalBuk7 days ago

Beautiful

Alan7 days ago

Bro please send me the full prompt

Bee7 days ago

Dm'ing..

Louis7 days ago

THIS IS DOPE! Is it back again to arena?

Bee7 days ago

I haven't been able to get any hits since today morning.. so not sure..

J A Z I I7 days ago

Ayoo the details are looking good

Bee7 days ago

Yeahh.. pretty good

Nafýr Gançiel Møfus7 days ago

i just wonder when they release this model

Bee7 days ago

Mid Oct probably

Kalevipoeg 🇪🇪7 days ago

wow!

The SI Therapist7 days ago

Gemini 4 Pro has fine taste now. I hope the next update is a pretty robot dog for my desk at home. Simple design rules make it easy to read too.

Bee7 days ago

Cool idea

有为huong7 days ago

这是真的吗

Bee7 days ago

Yes

有为huong7 days ago

什么时候能正式更新推送呢

Bee7 days ago

Probably mid October

Nivek .T7 days ago

The quality looks amazing!

Meir7 days ago

bro, this is like opus 5.5 level thing. so excited!

¢я~ѕαнιℓ7 days ago

Yeah the UI and design seems much better to me now.

ShadowAguy7 days ago

gemini 4 pro can make a demo look designed. porcelain bee, one tiny button.

Alex7 days ago

Bro that's so sleek. I wish they don't nerf it.

Wichtel7 days ago

Oh, it looks quiet dangerous.

Related Videos

Gemini 4 Pro surprises again! The prompt was to it to create a beautiful mechanical butterfly in three.js and needless to say the output is indeed beautiful! Earlier it was being tested under the placeholder "gemini-3.7-flash" but it seems to be have been removed from arena since..

Gemini 4 Pro surprises again! The prompt was to it to create a beautiful mechanical butterfly in three.js and needless to say the output is indeed beautiful! Earlier it was being tested under the placeholder "gemini-3.7-flash" but it seems to be have been removed from arena since..

Bee

606,328 views • 18 days ago

🚨 Gemini 4 Pro (Colosseum) This is from the third checkpoint in Arena rn, the style has massively improved since checkpoint 1 imo, this feels a lot more polished This is a pain to get in arena so I will post more examples when I get them, I will try to get comparisons to previous checkpoints / other models too

🚨 Gemini 4 Pro (Colosseum) This is from the third checkpoint in Arena rn, the style has massively improved since checkpoint 1 imo, this feels a lot more polished This is a pain to get in arena so I will post more examples when I get them, I will try to get comparisons to previous checkpoints / other models too

Lumina

98,381 views • 8 days ago

A comparison of the current and previous checkpoint of Gemini 4 Pro which is being tested on Arena under the name "gemini-3.8-flash".. The difference is clear, it has definitely improved a lot and there are yet more checkpoints to come! I don't like the UIs that it makes.. I hope that gets addressed!

A comparison of the current and previous checkpoint of Gemini 4 Pro which is being tested on Arena under the name "gemini-3.8-flash".. The difference is clear, it has definitely improved a lot and there are yet more checkpoints to come! I don't like the UIs that it makes.. I hope that gets addressed!

Bee

14,747 views • 12 days ago

Opus 4.5 (new model from Anthropic) can do something no model could do until now: It can close the design loop. This is (in my testing) the first model that can reliably do the visual loop of Generate Code ⇛ Render ⇛ Look at it ⇛ Improve. Step 3 (Visual critique) has been a real problem for so long. Models will be myopic or defensive when reviewing designs - it's been hard to tell if this has been a training or a vision problem. Opus 4.5 seems to have fixed it - it's not amazing, but it finally *can* do it. Claude models have historically (once again, in my testing) lagged behind in terms of visual understanding. Gemini for the last year has been the best - but there might be a new leader.

Opus 4.5 (new model from Anthropic) can do something no model could do until now: It can close the design loop. This is (in my testing) the first model that can reliably do the visual loop of Generate Code ⇛ Render ⇛ Look at it ⇛ Improve. Step 3 (Visual critique) has been a real problem for so long. Models will be myopic or defensive when reviewing designs - it's been hard to tell if this has been a training or a vision problem. Opus 4.5 seems to have fixed it - it's not amazing, but it finally can do it. Claude models have historically (once again, in my testing) lagged behind in terms of visual understanding. Gemini for the last year has been the best - but there might be a new leader.

Hrishi

58,331 views • 10 months ago

Seeing as Claude Mythos is releasing soon, I have two VERY astonishing outputs to share from it. 👀 ZERO-SHOT and LOW effort as well! These are the best outputs I've seen for this prompt ever since the October 2025 Gemini A/B models.

Seeing as Claude Mythos is releasing soon, I have two VERY astonishing outputs to share from it. 👀 ZERO-SHOT and LOW effort as well! These are the best outputs I've seen for this prompt ever since the October 2025 Gemini A/B models.

Lentils

587,072 views • 4 months ago

🚨 Major Scoop: OpenAI has just expanded internal testing of GPT Astra, codenamed "mozaik-alpha-fdm", signifying release might be near AND of course like previous times, we have the FIRST EVER public outputs of it for y'all😉 Both outputs are zero-shot on Max effort. Frontend *might* have finally been fixed, and the model gives a lot of attention to details

🚨 Major Scoop: OpenAI has just expanded internal testing of GPT Astra, codenamed "mozaik-alpha-fdm", signifying release might be near AND of course like previous times, we have the FIRST EVER public outputs of it for y'all😉 Both outputs are zero-shot on Max effort. Frontend might have finally been fixed, and the model gives a lot of attention to details

Lentils

764,948 views • 1 month ago

It would not be an exaggeration to say that Knights and Merchants is a cult game in many circles 🤔 It has been so long since I last played it that I have a hard time understanding what is going on. Although I do have some flashes from the past 😎🔊 🕹️Knights and Merchants,1998

It would not be an exaggeration to say that Knights and Merchants is a cult game in many circles 🤔 It has been so long since I last played it that I have a hard time understanding what is going on. Although I do have some flashes from the past 😎🔊 🕹️Knights and Merchants,1998

PeteZach

22,608 views • 8 months ago

It's been almost a week since the ICJ ruling on Gaza, and the government has still not said whether it accepts the court's authority and orders, and whether it will urge Israel to do the same. I asked the Attorney General again today for that assurance, but sadly did not get it.

It's been almost a week since the ICJ ruling on Gaza, and the government has still not said whether it accepts the court's authority and orders, and whether it will urge Israel to do the same. I asked the Attorney General again today for that assurance, but sadly did not get it.

Emily Thornberry

445,152 views • 2 years ago

Just recently found out that the demo for #Onimusha Way of the Sword got expanded, and it even has a special challenge boss. finally managed to beat it and not sure if we can call this No Damage since I casually blocked some attacks😅 The combat and animation work in this game are so good. It reminds me a lot of Sekiro, despite the differences between the two.

Just recently found out that the demo for #Onimusha Way of the Sword got expanded, and it even has a special challenge boss. finally managed to beat it and not sure if we can call this No Damage since I casually blocked some attacks😅 The combat and animation work in this game are so good. It reminds me a lot of Sekiro, despite the differences between the two.

BenchmarKing

12,722 views • 1 month ago

Brandi Glanville says she has finally received a diagnosis for her facial disfigurement after previously believing it might be a parasite. She has not revealed the diagnosis yet but says she will be sharing it soon. “I have a long road ahead of me to get fully better, but I finally have an answer, so I am just thrilled.” She adds that the entire situation has been hell and has turned her into an introvert, and she now feels anxiety when leaving the house. Source: TMZ

Brandi Glanville says she has finally received a diagnosis for her facial disfigurement after previously believing it might be a parasite. She has not revealed the diagnosis yet but says she will be sharing it soon. “I have a long road ahead of me to get fully better, but I finally have an answer, so I am just thrilled.” She adds that the entire situation has been hell and has turned her into an introvert, and she now feels anxiety when leaving the house. Source: TMZ

Bye Wig Hello Drama

906,999 views • 10 months ago

(I'M NOT CRYING😭😭😭😭) 🦈: Now I can say that i am happy to have a partner and even happier that it is William. Because if it were someone else and they encountered this kind of problems and stress, they might not have been a couple(he meant anyone else might not stay together). It is because of William that we are still a couple. WILLIAMEST PRESSTOUR #WilliamEstFanconPressTour #Flex1045xWilliamEstFANCON #WilliamEst

(I'M NOT CRYING😭😭😭😭) 🦈: Now I can say that i am happy to have a partner and even happier that it is William. Because if it were someone else and they encountered this kind of problems and stress, they might not have been a couple(he meant anyone else might not stay together). It is because of William that we are still a couple. WILLIAMEST PRESSTOUR #WilliamEstFanconPressTour #Flex1045xWilliamEstFANCON #WilliamEst

WHAT IFF Marz_marz (Westie)

37,705 views • 1 year ago

It will be six years since I joined as Assistant Professor, and eight years since the inception of StochLab @ IISc. Quadrupeds have been the focus so far. Sharing a video snippet below. We will be open-sourcing some of our design and software for others to use. Stay tuned!

It will be six years since I joined as Assistant Professor, and eight years since the inception of StochLab @ IISc. Quadrupeds have been the focus so far. Sharing a video snippet below. We will be open-sourcing some of our design and software for others to use. Stay tuned!

Shishir Kolathaya

13,446 views • 8 months ago

Elon Musk: For Starship to work, we still need to solve a problem that has never been solved by anyone. “What's the single biggest remaining problem for Starship? It's having the heat shield be reusable. No one has ever made a reusable orbital heat shield. The heat shield's got to make it through the ascent phase without shocking a bunch of tiles. And then it's going to come back in and also not lose a bunch of tiles or overheat the main airframe. Isn't that hard because it's kind of fundamentally a consumable. It just needs to last a very long time. We have brought the ship back and had it do a soft landing in the ocean. I've done that a few times. But it lost a lot of tiles. It was not reusable without a lot of work. So even though it did come to a soft landing, it would not have been reusable without a lot of work. So it's not really reusable in that sense. That's the biggest problem that remains. So if you want to be able to land it, refill propellant, and fly again, you can't do this laborious inspection of 40,000 tiles type of thing.” Interview with John Collison and Dwarkesh Patel, February 5, 2026

Elon Musk: For Starship to work, we still need to solve a problem that has never been solved by anyone. “What's the single biggest remaining problem for Starship? It's having the heat shield be reusable. No one has ever made a reusable orbital heat shield. The heat shield's got to make it through the ascent phase without shocking a bunch of tiles. And then it's going to come back in and also not lose a bunch of tiles or overheat the main airframe. Isn't that hard because it's kind of fundamentally a consumable. It just needs to last a very long time. We have brought the ship back and had it do a soft landing in the ocean. I've done that a few times. But it lost a lot of tiles. It was not reusable without a lot of work. So even though it did come to a soft landing, it would not have been reusable without a lot of work. So it's not really reusable in that sense. That's the biggest problem that remains. So if you want to be able to land it, refill propellant, and fly again, you can't do this laborious inspection of 40,000 tiles type of thing.” Interview with John Collison and Dwarkesh Patel, February 5, 2026

ELON CLIPS

41,196 views • 7 months ago

Success! - US Driver Monitoring for EU cars OTA now available (FSD v14) - Validated with 2026.2.9.3 - Improved our backend and made it more secure, please use passkeys for access! - All the coding for the BE and Embedded was done on the Neo, it is a very nice handy tool! I can't wait for all the users to experience the massive difference in monitoring! To all the haters who mass-reported the main Tesla Android account - I'll get it back soon (Elon time). If you know anyone at X - please reach out, don't be a stranger! I don't care that much, but that account took 5 years to build and I have a lot of great content there. I survived a few DDoS attacks today, people have been also blocking accounts with incorrect passwords. Fixed all that, came back stronger and will keep pushing. Short drive demo attached, I'll make you listen to some Rammstein since it's 4 AM in Europe ;) PS: Store has been flooded, lead times will increase or I'll temporary suspend them until I manage to ship out most of current backlog (let me know your preference!)

Success! - US Driver Monitoring for EU cars OTA now available (FSD v14) - Validated with 2026.2.9.3 - Improved our backend and made it more secure, please use passkeys for access! - All the coding for the BE and Embedded was done on the Neo, it is a very nice handy tool! I can't wait for all the users to experience the massive difference in monitoring! To all the haters who mass-reported the main Tesla Android account - I'll get it back soon (Elon time). If you know anyone at X - please reach out, don't be a stranger! I don't care that much, but that account took 5 years to build and I have a lot of great content there. I survived a few DDoS attacks today, people have been also blocking accounts with incorrect passwords. Fixed all that, came back stronger and will keep pushing. Short drive demo attached, I'll make you listen to some Rammstein since it's 4 AM in Europe ;) PS: Store has been flooded, lead times will increase or I'll temporary suspend them until I manage to ship out most of current backlog (let me know your preference!)

Michał Gapiński

31,427 views • 6 months ago

We play 5 on 4, defense has to get 3 stops in a row to get off. If we can rotate, rebound and get stops 5 on 4 it should be a whole lot easier when it gets to 5 on 5

We play 5 on 4, defense has to get 3 stops in a row to get off. If we can rotate, rebound and get stops 5 on 4 it should be a whole lot easier when it gets to 5 on 5

Dawson Dickson

111,465 views • 1 year ago

Kyle Shanahan was asked if there are any regrets he may have with how the Brandon Aiyuk and 49ers situation unfolded: “I didn’t get to hang out with Aiyuk much since he tore his ACL in week 7 vs Kansas City. Been trying to get him back to our team since then and haven’t been able to pull that off, so haven’t had the opportunity to get him to be a part of our team. He’s been extremely distant since he got hurt. Was hoping it would happen once he got healthy, but it never happened.”

Kyle Shanahan was asked if there are any regrets he may have with how the Brandon Aiyuk and 49ers situation unfolded: “I didn’t get to hang out with Aiyuk much since he tore his ACL in week 7 vs Kansas City. Been trying to get him back to our team since then and haven’t been able to pull that off, so haven’t had the opportunity to get him to be a part of our team. He’s been extremely distant since he got hurt. Was hoping it would happen once he got healthy, but it never happened.”

Coach Yac 🗣

910,821 views • 9 months ago

🎤 Have you seen New giving his review? 🫍 I saw it, I saw it. This one has to go to him, since he's a dessert expert. Both of us has our own expertise, mine is savory food review. When it comes to dessert, I can only judge basically a 1 or a 10. 1 for not good, 10 for good. I can’t tell the difference between something that’s a 2, 3, 4, 5, 6, or 7. As for New, he can tell the difference whether it's a 5 or a 6. It’s kind of like me with curry and chili paste. I can take just a little taste and can immediately tell if it's a 5 or a 6 but when it comes to desserts, I can only say, “This is good” or “This is not good.” and that's it. 🎤 But that means you won't get a taste of the original since they're improving the drink. 🫍 Yeah, I saw it as well. I think that’s a good thing. Since we're a family, and when it comes to the areas we’re each experts in, we can share our ideas and contribute to make it better in the best way possible.

🎤 Have you seen New giving his review? 🫍 I saw it, I saw it. This one has to go to him, since he's a dessert expert. Both of us has our own expertise, mine is savory food review. When it comes to dessert, I can only judge basically a 1 or a 10. 1 for not good, 10 for good. I can’t tell the difference between something that’s a 2, 3, 4, 5, 6, or 7. As for New, he can tell the difference whether it's a 5 or a 6. It’s kind of like me with curry and chili paste. I can take just a little taste and can immediately tell if it's a 5 or a 6 but when it comes to desserts, I can only say, “This is good” or “This is not good.” and that's it. 🎤 But that means you won't get a taste of the original since they're improving the drink. 🫍 Yeah, I saw it as well. I think that’s a good thing. Since we're a family, and when it comes to the areas we’re each experts in, we can share our ideas and contribute to make it better in the best way possible.

deeコ

35,616 views • 1 month ago

MISSING THE PERFECT DONUT We haven't seen a PERFECTLY well performed donut since this moment... The last one to do it messed it up 😤 I need him to win a few more so we can see it again and again

MISSING THE PERFECT DONUT We haven't seen a PERFECTLY well performed donut since this moment... The last one to do it messed it up 😤 I need him to win a few more so we can see it again and again

Tirii

140,167 views • 5 months ago

Last workout of the year! As the year comes to an end, I’ve been checking a few baselines to see where I’m at and plan the year ahead. Training’s been a bit inconsistent recently, but with a reduced playing schedule coming up it’ll allow me to train a bit harder and experiment with some things I’ve always wanted to try. It can sometimes get boring over the years if you’re doing similar things relentlessly, year on year. Now that I’ve got a bit more freedom, I’ll be taking advantage of that and adding some fun back in. I’ll be jumping back on the speed train soon too, as my clubhead speed has been down since breaking my ankle last year. I’ll be sharing some of that in the new year and seeing what I can get it back to.

Last workout of the year! As the year comes to an end, I’ve been checking a few baselines to see where I’m at and plan the year ahead. Training’s been a bit inconsistent recently, but with a reduced playing schedule coming up it’ll allow me to train a bit harder and experiment with some things I’ve always wanted to try. It can sometimes get boring over the years if you’re doing similar things relentlessly, year on year. Now that I’ve got a bit more freedom, I’ll be taking advantage of that and adding some fun back in. I’ll be jumping back on the speed train soon too, as my clubhead speed has been down since breaking my ankle last year. I’ll be sharing some of that in the new year and seeing what I can get it back to.

Oliver Wilson

15,748 views • 9 months ago