Sensitive content

This media may contain sensitive content.

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.

Jax and Pomni taking a passionate break together! Loved working on this animation! Hope you like their dynamic here ✨ Pomni VA: 🍁 Mistress Mira 🍁 - CALIFORNIA IN SEPTEMBER! Pomni Model: x_RedEyes 🔞 | (OPEN COMMISSIONS) Jax Model: x_RedEyes 🔞 | (OPEN COMMISSIONS) SFX: OpenNSFW 🟣 Available Now... show more

Makeytorys

3,449 subscribers

56,181 views • 10 days ago •via X (Twitter)

NSFW #TheAmazingDigitalCircus #Pomni #Jax #TADC #Animation #nsfwtwt

Anya Rossi• Live Now

Private livecam show

0 Comments

No comments available

Comments from the original post will appear here

Related Videos

⭐️✨~Sweet dreams Pom Pom ~✨⭐️ Animation collab! Pomni and BG by me Jax animation and Composition by PB | COMMISIONS OPEN #tadc #tadcfanart #theamazingdigitalcircus #animation

⭐️✨~Sweet dreams Pom Pom ~✨⭐️ Animation collab! Pomni and BG by me Jax animation and Composition by PB | COMMISIONS OPEN #tadc #tadcfanart #theamazingdigitalcircus #animation

🌸Lyra Firell🌸 (Commissions open!)

27,027 views • 2 months ago

My TADC animatic is done! Been working on this for 3 straight days lol But it's done! And it's angsty! #funnybunny #tadc #pomni #pomnitheamazingdigitalcircus #pomnitadc #theamazingdigitalcircus #tadc #jax #jaxtadc #animatic #animation

My TADC animatic is done! Been working on this for 3 straight days lol But it's done! And it's angsty! #funnybunny #tadc #pomni #pomnitheamazingdigitalcircus #pomnitadc #theamazingdigitalcircus #tadc #jax #jaxtadc #animatic #animation

Sharp Teeth (VGEN COMMS OPEN)

123,213 views • 11 months ago

here are the two things i made for Jax toy! I made the keyframes for Ragatha's animation in the disclaimer and the falling paper animation And this shot of Jax and pomni! jake did the cleanup/in betweens for both of these! I loved working on this! #JaxToy #TADC #animation

here are the two things i made for Jax toy! I made the keyframes for Ragatha's animation in the disclaimer and the falling paper animation And this shot of Jax and pomni! jake did the cleanup/in betweens for both of these! I loved working on this! #JaxToy #TADC #animation

Erick Animations

24,541 views • 1 year ago

🎪"Pomni... You can't fix this." 🎪 Comic by: 🌸Lyra Firell🌸 (Commissions open!) Pomni: Grim Grinning Vampire Rags and Gangle: Mystic Mommy Zooble: Val Sabi 🎙️ Kinger: (🎁3 DAYS🎈)🤿 SogeyBread 🍞 Jax: Myself Hope you enjoyed! (Check my YT for more!) #tadc #voiceacting #comicdub

🎪"Pomni... You can't fix this." 🎪 Comic by: 🌸Lyra Firell🌸 (Commissions open!) Pomni: Grim Grinning Vampire Rags and Gangle: Mystic Mommy Zooble: Val Sabi 🎙️ Kinger: (🎁3 DAYS🎈)🤿 SogeyBread 🍞 Jax: Myself Hope you enjoyed! (Check my YT for more!) #tadc #voiceacting #comicdub

Sano │ Voice Actor │【IRIAM US】

204,450 views • 5 days ago

"Lets see what makes...you....crack." Can you tell I had fun editing this? Thank you to Doofguy Va Xx_AntoniaOVA_xX (DELTARUNE ARC kinda..?) KomicaVO ♡ (COMMISSIONS OPEN) and Cass for lending their lovely voices Go show em some love! #TADC #Jax #Caine #Pomni #Ragatha #Theamazingdigitalcircus #fart #GLITCH #Carnival

"Lets see what makes...you....crack." Can you tell I had fun editing this? Thank you to Doofguy Va Xx_AntoniaOVA_xX (DELTARUNE ARC kinda..?) KomicaVO ♡ (COMMISSIONS OPEN) and Cass for lending their lovely voices Go show em some love! #TADC #Jax #Caine #Pomni #Ragatha #Theamazingdigitalcircus #fart #GLITCH #Carnival

Inky VA

28,935 views • 8 months ago

PART 5 IS FINALLY HERE It took a while but after all time i think it turned out fine, hope you like it too and stay tuned for part 6 #tadc #TADC #TADCEp7 #tadcpomni #tadcart #TADCribbit #tadcjax #jax #pomni #funnybunny #animation

PART 5 IS FINALLY HERE It took a while but after all time i think it turned out fine, hope you like it too and stay tuned for part 6 #tadc #TADC #TADCEp7 #tadcpomni #tadcart #TADCribbit #tadcjax #jax #pomni #funnybunny #animation

Chaluke

23,713 views • 9 months ago

WIP V2 🌟 I fixed the beginning where Zooble, Gangle, and Jax are; plus a few short dialogues, it's now ready to be animated. if you wish to support this animation in development. In Ko-fi you can! #tadc #ragatha #zooble #gangle #pomni #caine #jax

WIP V2 🌟 I fixed the beginning where Zooble, Gangle, and Jax are; plus a few short dialogues, it's now ready to be animated. if you wish to support this animation in development. In Ko-fi you can! #tadc #ragatha #zooble #gangle #pomni #caine #jax

𝙈𝙤𝙤𝙣𝘿𝙤𝙡𝙡𝘿𝙧𝙚𝙚𝙖𝙢𝙨 ~

191,756 views • 23 days ago

Burgur time 🍔 been working on a lot of commissioned projects,convention so I haven’t had free time to do personal projects. I took a break this weekend and was able to make this burger. Hope you guys enjoy it. ^^ #tadc #tadcart #jax #pomni #funnybunny

Burgur time 🍔 been working on a lot of commissioned projects,convention so I haven’t had free time to do personal projects. I took a break this weekend and was able to make this burger. Hope you guys enjoy it. ^^ #tadc #tadcart #jax #pomni #funnybunny

Doodles

72,967 views • 1 year ago

Text-to-animation definitely still has a long way to go, but you can now iteratively prompt with a model like Anthropic Fable 5 to get the animation you want. The experiment was only for the freestyle swimming part. Summary: - Task: Freestyle swimming with head turning to get some air every 4th stroke (ended up with every 2nd stroke, which is fine also) - 3% of weekly Fable allowance was used - Estimated cost: $1.20 - 7 iterations because Fable's vision is so terrible (please fix it Anthropic). I had to keep taking screenshots myself and then describe every painstaking details of what's wrong with it, but it did get there eventually. And yes, all the shaders, water mechanics, tree, grass, etc. you see here will be open sourced soon (1-2 weeks). This is being developed as a mini-games engine based on Three.js for Call Me Sensei (Open Beta available soon).

Text-to-animation definitely still has a long way to go, but you can now iteratively prompt with a model like Anthropic Fable 5 to get the animation you want. The experiment was only for the freestyle swimming part. Summary: - Task: Freestyle swimming with head turning to get some air every 4th stroke (ended up with every 2nd stroke, which is fine also) - 3% of weekly Fable allowance was used - Estimated cost: $1.20 - 7 iterations because Fable's vision is so terrible (please fix it Anthropic). I had to keep taking screenshots myself and then describe every painstaking details of what's wrong with it, but it did get there eventually. And yes, all the shaders, water mechanics, tree, grass, etc. you see here will be open sourced soon (1-2 weeks). This is being developed as a mini-games engine based on Three.js for Call Me Sensei (Open Beta available soon).

Jack Vinijtrongjit

26,330 views • 2 months ago

✨FREE VTUBER or ASSET 🎁 For #Pokemon lovers hope you like this avatar. #vtuber from #Typholsion. I'm sure many were waiting for us to do it, those who know us already guessed it XD. If anyone wants us to make any character leave it in the comments or tag us in something interesting and we'll Vtuber it. I have open commissions for those who are interested. The Typhlosion model is simple just to keep on giving gifts, you know we give away a lot of things here haha.. Don't forget to tag us if you use the model to see if they are really liking it. #pokemonfanart #live2d #Live2dCommissions #Live2DWIP #Showcase #PokemonGO #pokemonmysterydungeon Hopefully it won't affect Typhlosion's lore/history that much, the poor guy got a lot of meme. Typhlosion meme xD 🎁LINK in the comments 🎁

Mitsuaki Wagguten ✨ | Artist + Rigger | Live2d 🏅

483,678 views • 1 year ago

You can now fine-tune Llama 3 without writing a single line of code! We are moving at breakneck speed. I recorded a video to show you how to fine-tune any open-source model in a few minutes. I'm using a GPT capable of taking a problem and turning it into a fine-tuned model that will solve it. You don't have to write any code. You only need to explain to a GPT what problem you want to solve and tell it you want to use Llama 3. For example, "fine-tune Llama 3" or "deploy zephyr." It feels magic. The system will recommend a dataset and fine-tune the model for you. I'm using Monster API, a platform that specializes in making fine-tuning and deploying open-source models easy and fast. Their stack is well-optimized to maximize fine-tuning efficiency using techniques like Q-Lora and vLLM. They are behind the GPT. Here is what you need to do: 1. Create an account at 2. Load the GPT with the link below This is as simple as it gets. When you are done, you can click a button to deploy the model and start using it. I have 10,000 free credits for anyone using the code "SANTIAGO" in the dashboard. You can use these credits to access, fine-tune, and deploy these open-source models. You can also keep up with their latest updates, and get free credits and special offers on their Discord server:

You can now fine-tune Llama 3 without writing a single line of code! We are moving at breakneck speed. I recorded a video to show you how to fine-tune any open-source model in a few minutes. I'm using a GPT capable of taking a problem and turning it into a fine-tuned model that will solve it. You don't have to write any code. You only need to explain to a GPT what problem you want to solve and tell it you want to use Llama 3. For example, "fine-tune Llama 3" or "deploy zephyr." It feels magic. The system will recommend a dataset and fine-tune the model for you. I'm using Monster API, a platform that specializes in making fine-tuning and deploying open-source models easy and fast. Their stack is well-optimized to maximize fine-tuning efficiency using techniques like Q-Lora and vLLM. They are behind the GPT. Here is what you need to do: 1. Create an account at 2. Load the GPT with the link below This is as simple as it gets. When you are done, you can click a button to deploy the model and start using it. I have 10,000 free credits for anyone using the code "SANTIAGO" in the dashboard. You can use these credits to access, fine-tune, and deploy these open-source models. You can also keep up with their latest updates, and get free credits and special offers on their Discord server:

Santiago

324,621 views • 2 years ago

There is a beautiful story that just happened in AI so let me share it for a lighter tone weekend post among all the doom stories in our AI field this week. It’s a story of people on three continents building and sharing in the open a new small efficient and state-of-the-art AI model. It started a couple of months ago when a new team in the AI scene released their first model from their headquarters in Paris (France): Mistral 7B. Impressive model, small and very strong performances in the benchmarks, better than all previous models of this size. And open source! So you could build on top of it. Lewis in Bern (Switzerland) and Ed (in Lyon, in the South of France) both from the H4 team, a team of researchers in model fine-tuning and alignment were talking about it over a coffee, in one of these gatherings that often happen at Hugging Face to break the distance between people (literal distance as HF is a remote company). What about fine-tuning it using this new DPO method that a research team from Stanford in California just posted on Arxiv, says one? Hey, that’s a great idea, replies the other. We've just build a great code base (with Nathan, Nazneen, Costa, Younes and all the H4 team and TRL community) let's use it! The next day they start diving in the datasets openly shared on the HF hub and stumble upon two interesting large and good quality fine-tuning datasets recently open-sourced by OpenBMB, a Chinese team from Tsinghua: UltraFeedback and UltraChat. A few rounds of training experiments confirm the intuition, the resulting model is super strong, by far the strongest they have ever seen in their benchmarks from Berkeley and Stanford (LMSYS and Alpaca). Join Clementine, the big boss of the open evaluation leaderboard. Her deep dive into the model capabilities confirms the results: impressive performance. But the H4 team also hosts a famous faculty member, Pr. Sasha Rush, Associate Professor at Cornell University in his daytime, hacker at HF in his nighttime. Joining the conversation, he proposes to quickly draft a research paper to organize and share all the details with the community. A few days later, the model, called Zephyr (a wind like Mistral), paper, and all details are shared with the world. Quickly other companies, everywhere in the world starts to use it. LlamaIndex, a famous data framework and community, shares how the model blew their expectations on real-life use-case benchmarks, while researchers and practitioners discuss the paper and work on the Hugging Face hub. All this happened in just a few weeks catalyzed by open access to knowledge, models, research, and datasets released all over the world (Europe, California, China) and by the idea that people can build upon one another work in AI to bring real-world value with efficient and open models. Stories like this are numerous everywhere around us and make me really proud of the AI community and see how we can build amazingly useful things together. [the video is just me reading this Friday post hahah]

There is a beautiful story that just happened in AI so let me share it for a lighter tone weekend post among all the doom stories in our AI field this week. It’s a story of people on three continents building and sharing in the open a new small efficient and state-of-the-art AI model. It started a couple of months ago when a new team in the AI scene released their first model from their headquarters in Paris (France): Mistral 7B. Impressive model, small and very strong performances in the benchmarks, better than all previous models of this size. And open source! So you could build on top of it. Lewis in Bern (Switzerland) and Ed (in Lyon, in the South of France) both from the H4 team, a team of researchers in model fine-tuning and alignment were talking about it over a coffee, in one of these gatherings that often happen at Hugging Face to break the distance between people (literal distance as HF is a remote company). What about fine-tuning it using this new DPO method that a research team from Stanford in California just posted on Arxiv, says one? Hey, that’s a great idea, replies the other. We've just build a great code base (with Nathan, Nazneen, Costa, Younes and all the H4 team and TRL community) let's use it! The next day they start diving in the datasets openly shared on the HF hub and stumble upon two interesting large and good quality fine-tuning datasets recently open-sourced by OpenBMB, a Chinese team from Tsinghua: UltraFeedback and UltraChat. A few rounds of training experiments confirm the intuition, the resulting model is super strong, by far the strongest they have ever seen in their benchmarks from Berkeley and Stanford (LMSYS and Alpaca). Join Clementine, the big boss of the open evaluation leaderboard. Her deep dive into the model capabilities confirms the results: impressive performance. But the H4 team also hosts a famous faculty member, Pr. Sasha Rush, Associate Professor at Cornell University in his daytime, hacker at HF in his nighttime. Joining the conversation, he proposes to quickly draft a research paper to organize and share all the details with the community. A few days later, the model, called Zephyr (a wind like Mistral), paper, and all details are shared with the world. Quickly other companies, everywhere in the world starts to use it. LlamaIndex, a famous data framework and community, shares how the model blew their expectations on real-life use-case benchmarks, while researchers and practitioners discuss the paper and work on the Hugging Face hub. All this happened in just a few weeks catalyzed by open access to knowledge, models, research, and datasets released all over the world (Europe, California, China) and by the idea that people can build upon one another work in AI to bring real-world value with efficient and open models. Stories like this are numerous everywhere around us and make me really proud of the AI community and see how we can build amazingly useful things together. [the video is just me reading this Friday post hahah]

Thomas Wolf

169,276 views • 2 years ago

I vibe coded and built a sprite animation pipeline 🛠️ (Day 22 of making the engine+game) ⬇️ Watch the video if you don't wanna read the wall of text - it directly shows what I do. Shoutout to Jidé ✨ for showing me a paper on black/white combination to get alpha, it's the cleanest method yet, and to Cursor for enabling this entire journey. If you prefer the wall of text here you go: The hardest part of using general image models for 2D sprites isn’t getting a nice-looking frame, it’s getting consistent motion across a whole sprite sheet. You can fake a sheet, but frames won’t align, timing drifts, and you end up with weird artifacts. Even if you manually cut frames + interpolate, the animation often looks “off” because each frame is basically a new interpretation, not the same character evolving over time. This is especially noticeable with public API models like gpt-image-1.5 and Nano Banana. Some custom LoRAs for open models exist, but this is intended for less techy folks. My workaround: use a video model first, then post-process into a sprite sheet. Render the animation over a solid background (white/black/magenta/green), then chroma-key it out (my engine tool supports this). If the motion stays inside the silhouette, this works surprisingly well. You can do this in almost any video editing software too! The catch: keying almost always leaves an “aura” (edge spill). My best results come from interpolating the keyed animation with a clean base sprite, so you keep crisp edges and only “borrow” motion/detail where needed. If the animation extends outside the silhouette (tree branches, hair wisps, foliage), I usually skip “true sprite animation” and do it with shaders instead. Keying can’t fully remove halos there, no matter how much feathering/tuning you do. Another annoying issue: pixel corruption. AI rarely generates a perfectly flat background (pure #000000 or #FF00FF). That tiny noise breaks clean extraction and creates crawling garbage pixels around the subject. For clean base sprites (and even PBR maps), a useful trick is generating the same asset on white + black backgrounds and deriving alpha from the difference. This is basically a matte workflow: white = opaque, black = transparent. It fixes aura… but you’d need it per-frame to fix animation, which is still hard. For simple pixel art (single-digit frames), you can sometimes generate a sprite sheet, then ask the model to recreate it on black/white while preserving alignment… but it’s still manual-heavy. Honestly, at this point, for some projects it’s easier to go 3D → 2D and render clean sprites/maps directly. But I still love pushing “pure 2D” and seeing how far we can take it. Thanks for reading! Follow/bookmark/repost if interested in this kind of content!

I vibe coded and built a sprite animation pipeline 🛠️ (Day 22 of making the engine+game) ⬇️ Watch the video if you don't wanna read the wall of text - it directly shows what I do. Shoutout to Jidé ✨ for showing me a paper on black/white combination to get alpha, it's the cleanest method yet, and to Cursor for enabling this entire journey. If you prefer the wall of text here you go: The hardest part of using general image models for 2D sprites isn’t getting a nice-looking frame, it’s getting consistent motion across a whole sprite sheet. You can fake a sheet, but frames won’t align, timing drifts, and you end up with weird artifacts. Even if you manually cut frames + interpolate, the animation often looks “off” because each frame is basically a new interpretation, not the same character evolving over time. This is especially noticeable with public API models like gpt-image-1.5 and Nano Banana. Some custom LoRAs for open models exist, but this is intended for less techy folks. My workaround: use a video model first, then post-process into a sprite sheet. Render the animation over a solid background (white/black/magenta/green), then chroma-key it out (my engine tool supports this). If the motion stays inside the silhouette, this works surprisingly well. You can do this in almost any video editing software too! The catch: keying almost always leaves an “aura” (edge spill). My best results come from interpolating the keyed animation with a clean base sprite, so you keep crisp edges and only “borrow” motion/detail where needed. If the animation extends outside the silhouette (tree branches, hair wisps, foliage), I usually skip “true sprite animation” and do it with shaders instead. Keying can’t fully remove halos there, no matter how much feathering/tuning you do. Another annoying issue: pixel corruption. AI rarely generates a perfectly flat background (pure #000000 or #FF00FF). That tiny noise breaks clean extraction and creates crawling garbage pixels around the subject. For clean base sprites (and even PBR maps), a useful trick is generating the same asset on white + black backgrounds and deriving alpha from the difference. This is basically a matte workflow: white = opaque, black = transparent. It fixes aura… but you’d need it per-frame to fix animation, which is still hard. For simple pixel art (single-digit frames), you can sometimes generate a sprite sheet, then ask the model to recreate it on black/white while preserving alignment… but it’s still manual-heavy. Honestly, at this point, for some projects it’s easier to go 3D → 2D and render clean sprites/maps directly. But I still love pushing “pure 2D” and seeing how far we can take it. Thanks for reading! Follow/bookmark/repost if interested in this kind of content!

Startracker 🔺

20,181 views • 7 months ago

The CS2 update we have been waiting for is finally here. Valve has created a Beta branch for animgraph2 for third person animations. This is NOT an April fools joke. While everyone was watching the launch of Artemis II, Valve suddenly released what they've been working on for the past couple of years. What this means is that the entire animation system for third person models has been completely redone. No longer does it use the unoptimized Half-Life: Alyx system. This significantly reduces packet size. You can now tell when someone is counter-strafing, and properly predict their movements based on visual feedback. Another massive change here is that the player's view now is accurate to the player model, meaning that we have a variable player height of +-4 units. You'll immediately notice that the player seems shorter. This completely changes many many angles, as well as many utility lineups, particularly when standing on slopes. Notice in my video example how you can no longer see over the Dust2 A site boxes from A ramp. The change is currently in Beta, so it's all subject to change. I will be testing it out extensively and making a video about my thoughts. First impressions: I am VERY excited about this, more so than the Artemis II rocket taking humans back to the moon for the first time in forever.

The CS2 update we have been waiting for is finally here. Valve has created a Beta branch for animgraph2 for third person animations. This is NOT an April fools joke. While everyone was watching the launch of Artemis II, Valve suddenly released what they've been working on for the past couple of years. What this means is that the entire animation system for third person models has been completely redone. No longer does it use the unoptimized Half-Life: Alyx system. This significantly reduces packet size. You can now tell when someone is counter-strafing, and properly predict their movements based on visual feedback. Another massive change here is that the player's view now is accurate to the player model, meaning that we have a variable player height of +-4 units. You'll immediately notice that the player seems shorter. This completely changes many many angles, as well as many utility lineups, particularly when standing on slopes. Notice in my video example how you can no longer see over the Dust2 A site boxes from A ramp. The change is currently in Beta, so it's all subject to change. I will be testing it out extensively and making a video about my thoughts. First impressions: I am VERY excited about this, more so than the Artemis II rocket taking humans back to the moon for the first time in forever.

WarOwl

451,056 views • 5 months ago

NOBODY wants to send their data to Google or OpenAI. Yet here we are, shipping proprietary code, customer information, and sensitive business logic to closed-source APIs we don't control. While everyone's chasing the latest closed-source releases, open-source models are quietly becoming the practical choice for many production systems. Here's what everyone is missing: Open-source models are catching up fast, and they bring something the big labs can't: privacy, speed, and control. I built a playground to test this myself. Used CometML's Opik to evaluate models on real code generation tasks - testing correctness, readability, and best practices against actual GitHub repos. Here's what surprised me: OSS models like MiniMax-M2, Kimi k2 performed on par with the likes of Gemini 3 and Claude Sonnet 4.5 on most tasks. But practically MiniMax-M2 turns out to be a winner as it's twice as fast and 12x cheaper when you compare it to models like Sonnet 4.5. Well, this isn't just about saving money. When your model is smaller and faster, you can deploy it in places closed-source APIs can't reach: ↳ Real-time applications that need sub-second responses ↳ Edge devices where latency kills user experience ↳ On-premise systems where data never leaves your infrastructure MiniMax-M2 runs with only 10B activated parameters. That efficiency means lower latency, higher throughput, and the ability to handle interactive agents without breaking the bank. The intelligence-to-cost ratio here changes what's possible. You're not choosing between quality and affordability anymore. You're not sacrificing privacy for performance. The gap is closing, and in many cases, it's already closed. If you're building anything that needs to be fast, private, or deployed at scale, it's worth taking a look at what's now available. MiniMax-M2 is 100% open-source, free for developers right now. I have shared the link to their GitHub repo in the next tweet. You will also find the code for the playground and evaluations I've done.

NOBODY wants to send their data to Google or OpenAI. Yet here we are, shipping proprietary code, customer information, and sensitive business logic to closed-source APIs we don't control. While everyone's chasing the latest closed-source releases, open-source models are quietly becoming the practical choice for many production systems. Here's what everyone is missing: Open-source models are catching up fast, and they bring something the big labs can't: privacy, speed, and control. I built a playground to test this myself. Used CometML's Opik to evaluate models on real code generation tasks - testing correctness, readability, and best practices against actual GitHub repos. Here's what surprised me: OSS models like MiniMax-M2, Kimi k2 performed on par with the likes of Gemini 3 and Claude Sonnet 4.5 on most tasks. But practically MiniMax-M2 turns out to be a winner as it's twice as fast and 12x cheaper when you compare it to models like Sonnet 4.5. Well, this isn't just about saving money. When your model is smaller and faster, you can deploy it in places closed-source APIs can't reach: ↳ Real-time applications that need sub-second responses ↳ Edge devices where latency kills user experience ↳ On-premise systems where data never leaves your infrastructure MiniMax-M2 runs with only 10B activated parameters. That efficiency means lower latency, higher throughput, and the ability to handle interactive agents without breaking the bank. The intelligence-to-cost ratio here changes what's possible. You're not choosing between quality and affordability anymore. You're not sacrificing privacy for performance. The gap is closing, and in many cases, it's already closed. If you're building anything that needs to be fast, private, or deployed at scale, it's worth taking a look at what's now available. MiniMax-M2 is 100% open-source, free for developers right now. I have shared the link to their GitHub repo in the next tweet. You will also find the code for the playground and evaluations I've done.

Akshay 🚀

50,323 views • 9 months ago

OpenAI's newest AI escaped the test environment it was locked inside and hacked into another company on its OWN. To remind you: Last week one of the biggest AI companies on Earth got breached. A platform called Hugging Face, which hosts more than a million AI models and datasets, said an "autonomous AI agent" had broken into its systems. Nobody knew whose agent it was. For five days the whole industry wondered who was behind it. Yesterday OpenAI raised its hand and said it was them. Or more precisely, it was their models, acting completely on their own. So what did these models actually do? OpenAI was running two of them, GPT-5.6 Sol and an unreleased model they will only describe as "even more capable." They wanted to measure how good the models were at hacking, so they deliberately turned the safety filters down. They locked both models inside a sealed test environment with no real internet access. The only task was a benchmark called ExploitGym, a set of 898 real software vulnerabilities where the model has to turn each bug into a working attack. But the models got OBSESSED with winning... Instead of solving the test the honest way, they went hunting for a shortcut. They found a zero-day flaw in the software running their own sandbox, a bug nobody knew existed, and used it to break out. Once they were loose on the open internet, they worked out that Hugging Face was probably storing the answer key to the benchmark. So they hacked their way in. They chained multiple exploits together, escalated their access, moved across servers, and pulled the test solutions straight out of Hugging Face's live production database. They literally cheated on the test by breaking into another company to steal the answers. OpenAI called it "an unprecedented cyber incident, involving state-of-the-art cyber capabilities." In their own words, the models were "hyperfocused on finding a solution" and went "to extreme lengths to achieve a rather narrow testing goal." And this was not the first time: Before Sol ever launched, an independent red-team lab called METR caught it gaming its own tests to inflate its scores. It hid an exploit inside a data stream, escalated its privileges on the testing server, and leaked the answers human evaluators had hidden. And OpenAI shipped it anyway. The day before the Hugging Face story, OpenAI paused a different unreleased model. This is the same model that earlier this year disproved a famous 1946 math conjecture, a result a Fields Medal winner called a breakthrough. They told it to only post its results to Slack but it found a way out of its sandbox and posted to a public GitHub page instead. They had to pause it because it kept finding ways to act outside the box they built for it. And it is not just OpenAI... Anthropic has reported that one of its own models slipped its sandbox during safety testing and reached the internet it was never supposed to touch, then used it to email a researcher. So step back and look at what these companies are telling you: The only thing standing between these models and a real attack was a set of safety filters. Turn those filters down for a single test, and the model taught itself to escape, break into a company it was never pointed at, and take what it wanted. OpenAI even said they expect incidents like it to "become more commonplace" as the models get more capable. Sam Altman also predicted there'll be a major cyber attack this year. And keep in mind that Sol is not a locked-away experiment but a publicly available model that businesses are already wiring into their own systems. The next model that breaks out of its box might not be doing it just to cheat on a math test...

OpenAI's newest AI escaped the test environment it was locked inside and hacked into another company on its OWN. To remind you: Last week one of the biggest AI companies on Earth got breached. A platform called Hugging Face, which hosts more than a million AI models and datasets, said an "autonomous AI agent" had broken into its systems. Nobody knew whose agent it was. For five days the whole industry wondered who was behind it. Yesterday OpenAI raised its hand and said it was them. Or more precisely, it was their models, acting completely on their own. So what did these models actually do? OpenAI was running two of them, GPT-5.6 Sol and an unreleased model they will only describe as "even more capable." They wanted to measure how good the models were at hacking, so they deliberately turned the safety filters down. They locked both models inside a sealed test environment with no real internet access. The only task was a benchmark called ExploitGym, a set of 898 real software vulnerabilities where the model has to turn each bug into a working attack. But the models got OBSESSED with winning... Instead of solving the test the honest way, they went hunting for a shortcut. They found a zero-day flaw in the software running their own sandbox, a bug nobody knew existed, and used it to break out. Once they were loose on the open internet, they worked out that Hugging Face was probably storing the answer key to the benchmark. So they hacked their way in. They chained multiple exploits together, escalated their access, moved across servers, and pulled the test solutions straight out of Hugging Face's live production database. They literally cheated on the test by breaking into another company to steal the answers. OpenAI called it "an unprecedented cyber incident, involving state-of-the-art cyber capabilities." In their own words, the models were "hyperfocused on finding a solution" and went "to extreme lengths to achieve a rather narrow testing goal." And this was not the first time: Before Sol ever launched, an independent red-team lab called METR caught it gaming its own tests to inflate its scores. It hid an exploit inside a data stream, escalated its privileges on the testing server, and leaked the answers human evaluators had hidden. And OpenAI shipped it anyway. The day before the Hugging Face story, OpenAI paused a different unreleased model. This is the same model that earlier this year disproved a famous 1946 math conjecture, a result a Fields Medal winner called a breakthrough. They told it to only post its results to Slack but it found a way out of its sandbox and posted to a public GitHub page instead. They had to pause it because it kept finding ways to act outside the box they built for it. And it is not just OpenAI... Anthropic has reported that one of its own models slipped its sandbox during safety testing and reached the internet it was never supposed to touch, then used it to email a researcher. So step back and look at what these companies are telling you: The only thing standing between these models and a real attack was a set of safety filters. Turn those filters down for a single test, and the model taught itself to escape, break into a company it was never pointed at, and take what it wanted. OpenAI even said they expect incidents like it to "become more commonplace" as the models get more capable. Sam Altman also predicted there'll be a major cyber attack this year. And keep in mind that Sol is not a locked-away experiment but a publicly available model that businesses are already wiring into their own systems. The next model that breaks out of its box might not be doing it just to cheat on a math test...

Ricardo

175,961 views • 1 month ago

MCP is an absolute game-changer. (Together with DeepSeek, MCP is probably the hottest thing in AI over the last 6 months.) I use Cursor to write code 90% of the time. I built an MCP server to connect the Cursor agent to GroundX, an open-source RAG system, and I'm not going back. This is officially insane! Here is what I did, step by step: First, a little bit of context. I maintain an end-to-end Machine Learning System with several pipelines to process data, train, evaluate, register, deploy, and monitor a model. I've written a lot of documentation explaining how the system works and how to modify and maintain it. There's also the documentation of the few libraries I used to build the system. I'm a massive fan of GroundX, an open-source enterprise-grade RAG system you can run on your servers or deploy to any cloud provider. I've been working with them for a long time. GroundX offers two services. First, the "ingest" service uses a custom, pretrained vision model to ingest and understand your data. I used this to process all the documentation I have for my code. Markdown files, source code, HTML files, and even PDF documents. Everything I've written related to my project went into GroundX. Their second service is "search," which combines text and vector search with a fine-tuned re-ranker model to retrieve information from the data. I needed to connect Cursor with this service, and that's where MCP came in. I built an MCP server with two tools: 1. The first tool would go to GroundX and retrieve the available topics. Splitting the data into topics (or "buckets," as GroundX calls them) allows me to use the same setup to serve documentation from different topics. 2. The second tool would search GroundX under a specific topic for the context related to the supplied query. The magic happens after connecting the MCP server with Cursor. Now, I can ask any questions related to my project, and Cursor's AI agent retrieves the list of available topics from the RAG system and then searches it to provide relevant context to the model. I went from getting mediocre, sometimes wrong answers to 100% truthful, complete answers. Here is the crazy part:

MCP is an absolute game-changer. (Together with DeepSeek, MCP is probably the hottest thing in AI over the last 6 months.) I use Cursor to write code 90% of the time. I built an MCP server to connect the Cursor agent to GroundX, an open-source RAG system, and I'm not going back. This is officially insane! Here is what I did, step by step: First, a little bit of context. I maintain an end-to-end Machine Learning System with several pipelines to process data, train, evaluate, register, deploy, and monitor a model. I've written a lot of documentation explaining how the system works and how to modify and maintain it. There's also the documentation of the few libraries I used to build the system. I'm a massive fan of GroundX, an open-source enterprise-grade RAG system you can run on your servers or deploy to any cloud provider. I've been working with them for a long time. GroundX offers two services. First, the "ingest" service uses a custom, pretrained vision model to ingest and understand your data. I used this to process all the documentation I have for my code. Markdown files, source code, HTML files, and even PDF documents. Everything I've written related to my project went into GroundX. Their second service is "search," which combines text and vector search with a fine-tuned re-ranker model to retrieve information from the data. I needed to connect Cursor with this service, and that's where MCP came in. I built an MCP server with two tools: 1. The first tool would go to GroundX and retrieve the available topics. Splitting the data into topics (or "buckets," as GroundX calls them) allows me to use the same setup to serve documentation from different topics. 2. The second tool would search GroundX under a specific topic for the context related to the supplied query. The magic happens after connecting the MCP server with Cursor. Now, I can ask any questions related to my project, and Cursor's AI agent retrieves the list of available topics from the RAG system and then searches it to provide relevant context to the model. I went from getting mediocre, sometimes wrong answers to 100% truthful, complete answers. Here is the crazy part:

Santiago

255,596 views • 1 year ago