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.

1/20 I trained two segmentation models using GPT-6 Astra Ultra as the orchestrator with no human labels supplied. My prompt was very lazy too as I had to catch a flight. RF-DETR-Seg-M predictions: Left: SAM3 training masks Right: Astra training masks Both videos held out.

Satya Mallick

14,594 subscribers

28,485 views • 1 day ago •via X (Twitter)

Science & Technology Education Health & Wellness

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

i fine-tuned SAM3 on 1,800 aerial drone images. the loss went down. then it started labeling cars as boats, tricycles, and people all at once, on the same object two things were wrong and neither was obvious 1. i was training the entire vision encoder on 1,800 samples freezing the encoder and only training the mask decoder fixed the hallucinations immediately 2. my training data had zero negative examples, every prompt always matched a mask adding 30% negative pairs where wrong prompts map to empty masks fixed the rest two decisions i made so early i forgot they were decisions

i fine-tuned SAM3 on 1,800 aerial drone images. the loss went down. then it started labeling cars as boats, tricycles, and people all at once, on the same object two things were wrong and neither was obvious 1. i was training the entire vision encoder on 1,800 samples freezing the encoder and only training the mask decoder fixed the hallucinations immediately 2. my training data had zero negative examples, every prompt always matched a mask adding 30% negative pairs where wrong prompts map to empty masks fixed the rest two decisions i made so early i forgot they were decisions

harpreet

10,136 views • 2 months ago

Using GPT-6 Astra I was able to quickly protorype a sparse grid allocator fluid simulation for VRChat using my specified architecture, but honestly as cool as it is its absolutely overkill for VRChat. Isosurface ray tracing could be done here, but its too complex to work with.

Using GPT-6 Astra I was able to quickly protorype a sparse grid allocator fluid simulation for VRChat using my specified architecture, but honestly as cool as it is its absolutely overkill for VRChat. Isosurface ray tracing could be done here, but its too complex to work with.

Mykhailo Moroz

27,827 views • 2 days ago

OpenAI just released GPT-6 Astra, and OpenAI President Greg Brockman is calling it a generational leap, saying he personally believes it could mark the arrival of AGI. Astra launched on Thursday, September 3rd, 2026, following July's GPT-5.6 solo release. Brockman told reporters, "For me personally, I do think we're there. Directionally, I believe this. I think we're on a path to AGI. The timing I think is kind of uncertain, but we are definitely on a path to AGI." OpenAI executives say Astra moves past the chatbot era, capable of handling end to end work on its own across browsers, spreadsheets, codebases, and enterprise apps without a human in the loop. The model was trained on over 100,000 GPUs at the Stargate site in Texas. Access starts through OpenAI's gated Daybreak Access cybersecurity program for enterprise customers, then rolls out to ChatGPT Plus, Pro, Business, and Enterprise, plus the API and Amazon Web Services over the coming days. Ahead of launch, two models appeared in Codex, GPT-6 Astra and GPT-6 Astra Aeon, with the latter believed to be a higher effort, long horizon variant built for autonomous multi agent work.

OpenAI just released GPT-6 Astra, and OpenAI President Greg Brockman is calling it a generational leap, saying he personally believes it could mark the arrival of AGI. Astra launched on Thursday, September 3rd, 2026, following July's GPT-5.6 solo release. Brockman told reporters, "For me personally, I do think we're there. Directionally, I believe this. I think we're on a path to AGI. The timing I think is kind of uncertain, but we are definitely on a path to AGI." OpenAI executives say Astra moves past the chatbot era, capable of handling end to end work on its own across browsers, spreadsheets, codebases, and enterprise apps without a human in the loop. The model was trained on over 100,000 GPUs at the Stargate site in Texas. Access starts through OpenAI's gated Daybreak Access cybersecurity program for enterprise customers, then rolls out to ChatGPT Plus, Pro, Business, and Enterprise, plus the API and Amazon Web Services over the coming days. Ahead of launch, two models appeared in Codex, GPT-6 Astra and GPT-6 Astra Aeon, with the latter believed to be a higher effort, long horizon variant built for autonomous multi agent work.

AGTP

20,156 views • 5 days ago

I used to love playing with trains when I was younger. Recently, I saw someone create a voxel-based train table using Fable 5.1, so I had to try something similar with GPT-6 Astra using three.js. It turned out absolutely amazing! You can actually play and interact with it directly on the website, (link in the thread) start & stop the trains, rotate around the entire table, zoom in & out, & explore all the tiny details. The level of detail is really mind-blowing! This whole experience with Astra feels incredibly magical. I have so many other things I’m building right now that I don’t think I have ever felt this excited about creating! Childhood ideas, random inspirations, games I always wanted to play but don't yet exist, things I once could only imagine, I can now turn them into working, interactive experiences in matter of hours! So now we know what true AGI feels like: imagination becoming reality almost as fast as you can describe it!

I used to love playing with trains when I was younger. Recently, I saw someone create a voxel-based train table using Fable 5.1, so I had to try something similar with GPT-6 Astra using three.js. It turned out absolutely amazing! You can actually play and interact with it directly on the website, (link in the thread) start & stop the trains, rotate around the entire table, zoom in & out, & explore all the tiny details. The level of detail is really mind-blowing! This whole experience with Astra feels incredibly magical. I have so many other things I’m building right now that I don’t think I have ever felt this excited about creating! Childhood ideas, random inspirations, games I always wanted to play but don't yet exist, things I once could only imagine, I can now turn them into working, interactive experiences in matter of hours! So now we know what true AGI feels like: imagination becoming reality almost as fast as you can describe it!

Derya Unutmaz, MD

115,264 views • 5 days ago

Hassan Whiteside went from training at the YMCA to signing a $98,000,000 deal with the Miami Heat “I was going to the YMCA to stay in shape. I'm over there lifting, running, playing pickup on the side. That was my training facility. The YMCA was my training facility” “I had a workout with the Grizzlies. I was in training camp with them, and I think me and Earl Clark were two of the last guys to get let go. And I remember going to Miami to work out, and Pat Riley personally came out on the court and worked me out. Hardest workout of my life” “I left the workout and went down to the G League. The Grizzlies' G League team sent me down there. I had a really good game. I think I had like 30 and 20 or something. My agent called me playing ‘Welcome to Miami.’ He was like, Miami wants to sign you to a four-year, non-guaranteed deal” “I ended up signing a one year, and then I signed a two year non-guaranteed deal. A lot of people don't know I wasn’t even on a guaranteed contract for a lot of the time I was with the Heat”

Hassan Whiteside went from training at the YMCA to signing a $98,000,000 deal with the Miami Heat “I was going to the YMCA to stay in shape. I'm over there lifting, running, playing pickup on the side. That was my training facility. The YMCA was my training facility” “I had a workout with the Grizzlies. I was in training camp with them, and I think me and Earl Clark were two of the last guys to get let go. And I remember going to Miami to work out, and Pat Riley personally came out on the court and worked me out. Hardest workout of my life” “I left the workout and went down to the G League. The Grizzlies' G League team sent me down there. I had a really good game. I think I had like 30 and 20 or something. My agent called me playing ‘Welcome to Miami.’ He was like, Miami wants to sign you to a four-year, non-guaranteed deal” “I ended up signing a one year, and then I signed a two year non-guaranteed deal. A lot of people don't know I wasn’t even on a guaranteed contract for a lot of the time I was with the Heat”

dank

79,384 views • 1 month ago

Can GPT-6 Astra actually create a 3D model like this? 🤔 I’ve been seeing people hype GPT-6 Astra looooot when it comes to 3D modeling, but can it really create 3D models with this level of detail? Wish i had access to try it,.. anyway, I created this one using Tripo as my lil submission to Tripo’s Companion Challenge where you have to create a 3D companion of your choice. I generated a bunch of different imaginary companions using GPT Image 2.0, from cute little dragons and humans to these adorable animals. After trying out a lot of different ideas, I finally settled on this panda in a bamboo forest look. I love how quickly you can go from having a random idea in your head to actually seeing it come to life, these days. That part is honestly so much fun. Now I just need to animate this little guy and turn it into an animated panda game or maybe 3D print it out haha #tripop2

Can GPT-6 Astra actually create a 3D model like this? 🤔 I’ve been seeing people hype GPT-6 Astra looooot when it comes to 3D modeling, but can it really create 3D models with this level of detail? Wish i had access to try it,.. anyway, I created this one using Tripo as my lil submission to Tripo’s Companion Challenge where you have to create a 3D companion of your choice. I generated a bunch of different imaginary companions using GPT Image 2.0, from cute little dragons and humans to these adorable animals. After trying out a lot of different ideas, I finally settled on this panda in a bamboo forest look. I love how quickly you can go from having a random idea in your head to actually seeing it come to life, these days. That part is honestly so much fun. Now I just need to animate this little guy and turn it into an animated panda game or maybe 3D print it out haha #tripop2

The Bugged Dev

154,188 views • 4 days ago

I am super excited to share this educational video that I had GPT-6 Astra create from a single prompt: “Create a 5 minute educational video about T cells,” cells that I had devoted most of my life studying! I also had Astra to use the Remotion plugin for the video generation. Astra also used Imagegen to create the visuals, it also suggested to use HeyGen for narration (have not used that before) and then assembled the entire educational video on its own one-shot, including what to say, how to explain it! The result is so incredibly well made that, narration, the animations are all so perfectly created that even with 35 years of experience studying T cells, I really don’t think I could have explained this any better myself! The quality is also better than anything I had seen before, Astra is far ahead of all models now! I’ve also been creating much longer and more advanced videos on T cells, immunology, and related topics, and I’m now planning to build an entire video lecture series and put it all on a dedicated website. I am just having the time of my life making these, it actually makes me emotional to be able to magically create these with few prompts!

I am super excited to share this educational video that I had GPT-6 Astra create from a single prompt: “Create a 5 minute educational video about T cells,” cells that I had devoted most of my life studying! I also had Astra to use the Remotion plugin for the video generation. Astra also used Imagegen to create the visuals, it also suggested to use HeyGen for narration (have not used that before) and then assembled the entire educational video on its own one-shot, including what to say, how to explain it! The result is so incredibly well made that, narration, the animations are all so perfectly created that even with 35 years of experience studying T cells, I really don’t think I could have explained this any better myself! The quality is also better than anything I had seen before, Astra is far ahead of all models now! I’ve also been creating much longer and more advanced videos on T cells, immunology, and related topics, and I’m now planning to build an entire video lecture series and put it all on a dedicated website. I am just having the time of my life making these, it actually makes me emotional to be able to magically create these with few prompts!

Derya Unutmaz, MD

335,127 views • 5 days ago

I’m now making a World of Warcraft inspired single-player RPG with some RuneScape elements for the browser with GPT-6 Astra. I literally had the idea while drinking my morning coffee and thought… while I have another Astra project to develop treatments for cancer is ongoing, why not start this too! 😅 It’s still a very early version, but it’s already actually playable! Combat works, you can chop down trees and gather resources, and I’m starting to build out the world, progression, quests, loot, item tables and other RPG systems. I know this will surprise some but I used to be a serious gamer, I just know this stuff hah! Maybe I’ll eventually turn it into an MMORPG so we can form rival Codex vs. Claude vs Grok guilds, but no cheating with the AI bots. 😂

I’m now making a World of Warcraft inspired single-player RPG with some RuneScape elements for the browser with GPT-6 Astra. I literally had the idea while drinking my morning coffee and thought… while I have another Astra project to develop treatments for cancer is ongoing, why not start this too! 😅 It’s still a very early version, but it’s already actually playable! Combat works, you can chop down trees and gather resources, and I’m starting to build out the world, progression, quests, loot, item tables and other RPG systems. I know this will surprise some but I used to be a serious gamer, I just know this stuff hah! Maybe I’ll eventually turn it into an MMORPG so we can form rival Codex vs. Claude vs Grok guilds, but no cheating with the AI bots. 😂

Derya Unutmaz, MD

22,086 views • 4 days ago

One of my childhood dreams was to one day ride aboard the USS Enterprise NCC-1701 from the original Star Trek series. I have now realized perhaps the next best thing: creating the USS Enterprise from scratch with GPT-6 Astra using a CAD program! The entire starship was designed directly inside CAD (AutoFusion), with 28 moving parts, the captain’s bridge, several interior areas, and a design that could potentially be 3D printed! The video below was also created by Astra of the actual CAD designed model of the starship. I’ll share a separate video showing it in much more detail, and I’ll put all the CAD files on GitHub as soon as I finish designing a few more parts. 😊 This feels so magical! As Sam Altman has said, using AI now feels like having a genie that grants your wishes! A childhood dream that once existed only in my imagination is now something I can actually design, build, explore, and perhaps soon even hold in my hands! Only stars are the limit!

One of my childhood dreams was to one day ride aboard the USS Enterprise NCC-1701 from the original Star Trek series. I have now realized perhaps the next best thing: creating the USS Enterprise from scratch with GPT-6 Astra using a CAD program! The entire starship was designed directly inside CAD (AutoFusion), with 28 moving parts, the captain’s bridge, several interior areas, and a design that could potentially be 3D printed! The video below was also created by Astra of the actual CAD designed model of the starship. I’ll share a separate video showing it in much more detail, and I’ll put all the CAD files on GitHub as soon as I finish designing a few more parts. 😊 This feels so magical! As Sam Altman has said, using AI now feels like having a genie that grants your wishes! A childhood dream that once existed only in my imagination is now something I can actually design, build, explore, and perhaps soon even hold in my hands! Only stars are the limit!

Derya Unutmaz, MD

13,231 views • 5 days ago

Someone gave GPT-6 and Fable 5.1 access to Canva. One smoked the other (it's not even close). Claude Fable 5.1 is on the left. GPT-6 Astra is on the right. The gap is not small. It comes down to one test (ARC-AGI 3). It drops a model into a game with no instructions and makes it work them out. Astra solved levels in fewer moves than you or I would, 96% of the time. Claude Opus 5 managed less than half of that. That's why everyone is shouting about AGI again. You can turn this on yourself in about two minutes: 1. Open the ChatGPT desktop app, switch to Work. 2. Go to Plugins, then Computer Use. Install it. 3. Switch on the server and skill toggles. 4. Allow Screen Recording and Accessibility. 5. Point it at one app and ask it do something. I wrote up the full Claude → Codex migration. Access it here → Repost ♻️ to help someone in your network.

Someone gave GPT-6 and Fable 5.1 access to Canva. One smoked the other (it's not even close). Claude Fable 5.1 is on the left. GPT-6 Astra is on the right. The gap is not small. It comes down to one test (ARC-AGI 3). It drops a model into a game with no instructions and makes it work them out. Astra solved levels in fewer moves than you or I would, 96% of the time. Claude Opus 5 managed less than half of that. That's why everyone is shouting about AGI again. You can turn this on yourself in about two minutes: 1. Open the ChatGPT desktop app, switch to Work. 2. Go to Plugins, then Computer Use. Install it. 3. Switch on the server and skill toggles. 4. Allow Screen Recording and Accessibility. 5. Point it at one app and ask it do something. I wrote up the full Claude → Codex migration. Access it here → Repost ♻️ to help someone in your network.

Charlie Hills

38,355 views • 2 days ago

meta just dropped a bomb. Muse Spark 1.1 is 86% cheaper than GPT-5.6 Sol and 91% cheaper than Fable 5 on output tokens ($4.25 vs $30 vs $50 per M) I thought the model was going to be horrendous, but it ended up being insanely fast, and very capable. front end design is VERY strong. Agentic capabilities seem great as it was able to spawn and even orchestrate other agents inside my app cnvs using the mcp and cli tool with zero instruction. we have soo many insanely good models right now.

meta just dropped a bomb. Muse Spark 1.1 is 86% cheaper than GPT-5.6 Sol and 91% cheaper than Fable 5 on output tokens ($4.25 vs $30 vs $50 per M) I thought the model was going to be horrendous, but it ended up being insanely fast, and very capable. front end design is VERY strong. Agentic capabilities seem great as it was able to spawn and even orchestrate other agents inside my app cnvs using the mcp and cli tool with zero instruction. we have soo many insanely good models right now.

Max Blade

88,802 views • 2 months ago

Michael Beasley says his family sent him to prep school with $40 and didn't call him for two months "I went to Laurinburg Prep by myself and I didn't have no cell phone and I ain't know how to use a pay phone. So I end up not talking to nobody in my family for like the first two months, three months. And them sent me down there with $40" "I pinched that $40 for like two weeks. After that, I had to rely on everybody else" "I was just scared as. My mother ain't call. I was mad as. If you know Laurinburg, you got to figure that out. We was in the middle of the hood, right by a cemetery"

Michael Beasley says his family sent him to prep school with $40 and didn't call him for two months "I went to Laurinburg Prep by myself and I didn't have no cell phone and I ain't know how to use a pay phone. So I end up not talking to nobody in my family for like the first two months, three months. And them sent me down there with $40" "I pinched that $40 for like two weeks. After that, I had to rely on everybody else" "I was just scared as. My mother ain't call. I was mad as. If you know Laurinburg, you got to figure that out. We was in the middle of the hood, right by a cemetery"

Arslan

144,725 views • 1 month ago

GPT-5.6 vs GPT-5.5 on my custom spaceship prompt. I gave both models the exact same custom prompt. This is also the same prompt I previously gave to Fable 5. For context, GPT-5.6 Pro worked for 87 minutes, while GPT-5.5 Extra High worked for 34 minutes and 42 seconds. As I’ve said before, based on great authority GPT-5.6 will be an incremental/soldi improvement over GPT-5.5, not a “Fable killer.” My rough expectation has been that it would trade blows with Fable 5 on some benchmarks, maybe win around half depending on the category, but not clearly surpass it overall. And again fable five will have bigger model smell, but this was expected. After testing this coding output, that view feels pretty accurate. GPT-5.6 is clearly better than GPT-5.5 in several visual areas. The lighting, shading, chairs, object details, and exterior of the spaceship looked noticeably stronger. The scene was also easier to test. I do want to give GPT-5.5 credit though. It built out the rooms much much better and the planets looked better than GPT-5.6’s. It was also interesting that both GPT-5.5 and GPT-5.6 produced better-looking planets than Fable 5 in this specific test. The downside with GPT-5.5 was stability. The game was much glitchier and harder to test compared to GPT-5.6. But when it comes to the core of the demo, which is the spaceship itself, Fable 5 still beat both models pretty comfortably. GPT-5.6 is impressive, but from this test, it looks exactly like what I expected which was a meaningful incremental improvement over GPT-5.5, at least for indie game demos, but not something that replaces Fable 5. In collaboration with Chetaslua

GPT-5.6 vs GPT-5.5 on my custom spaceship prompt. I gave both models the exact same custom prompt. This is also the same prompt I previously gave to Fable 5. For context, GPT-5.6 Pro worked for 87 minutes, while GPT-5.5 Extra High worked for 34 minutes and 42 seconds. As I’ve said before, based on great authority GPT-5.6 will be an incremental/soldi improvement over GPT-5.5, not a “Fable killer.” My rough expectation has been that it would trade blows with Fable 5 on some benchmarks, maybe win around half depending on the category, but not clearly surpass it overall. And again fable five will have bigger model smell, but this was expected. After testing this coding output, that view feels pretty accurate. GPT-5.6 is clearly better than GPT-5.5 in several visual areas. The lighting, shading, chairs, object details, and exterior of the spaceship looked noticeably stronger. The scene was also easier to test. I do want to give GPT-5.5 credit though. It built out the rooms much much better and the planets looked better than GPT-5.6’s. It was also interesting that both GPT-5.5 and GPT-5.6 produced better-looking planets than Fable 5 in this specific test. The downside with GPT-5.5 was stability. The game was much glitchier and harder to test compared to GPT-5.6. But when it comes to the core of the demo, which is the spaceship itself, Fable 5 still beat both models pretty comfortably. GPT-5.6 is impressive, but from this test, it looks exactly like what I expected which was a meaningful incremental improvement over GPT-5.5, at least for indie game demos, but not something that replaces Fable 5. In collaboration with Chetaslua

Chris

250,919 views • 2 months ago

This woman was 12 minutes late to the gate and the airline gave their seats to other customers with standby tickets. The rules are rules for a reason but if the gates are still open, is this just being petty? They are there now so I don’t see why they can’t just have the seats they paid for. I remember when I was almost sprinting to the gate because the watch j was wearing stopped working, I just didn’t know it. I thought I had plenty of time so as I walked to get something to eat I saw some of the gate time displays and in the corner of the screen displayed a time. I was confused because when I looked at my watch it said 1:30pm, my flight left at 2:30pm and the time on the screen was 2:25pm. I kept walking and passed a gift shop that had a clock and it also said 2:25pm. I hurried to the employee at the shop and asked what time it was and they pointed to the clock meaning that was the actual time. I took off to the gate because I knew there was a big chance I wasn’t going to make my flight. I got there right at 2:30pm. They were getting just letting the last people on. I had my ticket in hand and she said she had to make sure I could still board. Initially when she walked back she had a worried look on her face as if she had bad news, I was prepared for the worst but luck was on my side as she said they agreed to let me on. My next flight I wasn’t so lucky but I will save that for another time. My luck ran out for the rest of the day though, it made me think if I was better off missing that flight. The guy I was sat by snored bad, I had a delay at my next connecting flight, and the food they gave me made me nauseated but I was still thankful I made it home on time.

This woman was 12 minutes late to the gate and the airline gave their seats to other customers with standby tickets. The rules are rules for a reason but if the gates are still open, is this just being petty? They are there now so I don’t see why they can’t just have the seats they paid for. I remember when I was almost sprinting to the gate because the watch j was wearing stopped working, I just didn’t know it. I thought I had plenty of time so as I walked to get something to eat I saw some of the gate time displays and in the corner of the screen displayed a time. I was confused because when I looked at my watch it said 1:30pm, my flight left at 2:30pm and the time on the screen was 2:25pm. I kept walking and passed a gift shop that had a clock and it also said 2:25pm. I hurried to the employee at the shop and asked what time it was and they pointed to the clock meaning that was the actual time. I took off to the gate because I knew there was a big chance I wasn’t going to make my flight. I got there right at 2:30pm. They were getting just letting the last people on. I had my ticket in hand and she said she had to make sure I could still board. Initially when she walked back she had a worried look on her face as if she had bad news, I was prepared for the worst but luck was on my side as she said they agreed to let me on. My next flight I wasn’t so lucky but I will save that for another time. My luck ran out for the rest of the day though, it made me think if I was better off missing that flight. The guy I was sat by snored bad, I had a delay at my next connecting flight, and the food they gave me made me nauseated but I was still thankful I made it home on time.

SonnyBoy🇺🇸

206,120 views • 1 month ago

J. R. Smith on when he was outside the Garden with the Knicks fans: “So, when I came out the Garden, bro my hotel was to the left and I looked to the right and the big a** mob was right there. And I was just like, man, I’m one with the people. So, I’m like man, let’s see what this about. As I’m walking over there, the closer I get, I keep realizing like everybody’s noticing that I’m coming over there or whatever. So, as every step I take it’s getting louder. It’s getting louder. I’m sitting there like, as I get closer, I’m like, “Yeah this probably not as good, bro. No bullsh*t dog.” It was, as I’m think that sh*t somebody 20 feet in front of me fall. And then that sh*t was a domino effect.” (Via Str8 to Da League)

J. R. Smith on when he was outside the Garden with the Knicks fans: “So, when I came out the Garden, bro my hotel was to the left and I looked to the right and the big a** mob was right there. And I was just like, man, I’m one with the people. So, I’m like man, let’s see what this about. As I’m walking over there, the closer I get, I keep realizing like everybody’s noticing that I’m coming over there or whatever. So, as every step I take it’s getting louder. It’s getting louder. I’m sitting there like, as I get closer, I’m like, “Yeah this probably not as good, bro. No bullsht dog.” It was, as I’m think that sht somebody 20 feet in front of me fall. And then that sh*t was a domino effect.” (Via Str8 to Da League)

NBA Courtside

412,403 views • 3 months ago