🚨 PREMIUM SUPPORT IN X IS A LIE A USER EXPERIENCE CASE, AND THIS IS NOT THE FIRST TIME 😭😭😭😭 @elonmusk @michaelnicollsx @X @premium @SpaceXSI 4 days trying to ask what is the problem with my account and they send me links to FAQ and tell me the premium support they advertise as an "advantage" of the plan is for monetized accounts only! The products here are like campaign promises, and the features they sell too. Everything is SUPER in the name only. As a premium plus subscriber of X I have less than 100 views per post, I reach out to "PREMIUM SUPPORT" and they give me links to FAQ page. When I ask if they can at least tell me if there is something wrong with my account they said: "Premium Support is for monetized accounts only" (or fully automated processes that never need any help) so why they sell it to everyone??? since Wednesday I am asking what is wrong with my account, why my posts don't reach 10% of my followers number (which is 600 only) in 48h? and I still couldn't get an answer

⚃ —— 🄷🄰🅂🄷🅃🄰🄶 —— ⚃ 【M】【4】【S】【T】【E】【R】 🟢🔶🔳🔻🟪🔵🟫 I decided that for fun and for “fan” (😂) And for the sake of bringing awareness about important names of modern art, illustrators and great character designers I will create a new series that will be flagged by the hashtag #M4STER 🩸 Most people think AI art is not Art, and that AI artists have no real clue about art. 🫵🏻 If you are one of these people I have to say: you are mostly completelly right! (😂😭😂😭😂) Most AI artists do not have a clue about art and do not have real understanding about the technology. They learned how to build glossy things and they try to use a medium made to be used in a way that ressambles more the role of a a movie director, as if they were painters. AI art in some situations can be also tied to a particular visual identity beyond prompting, but in many ways everything digital can be copied. The common sense (unfortunately) still looks to AI art and tries to make is fit in the existing frameworks of other known mediums. But AI is not like other mediums. Everybody can draw, paint and tske a photo. However not all of them are artists. Among the artists very few are recognized. Recognition requires talent and some resourcefulness. As in everything in today's world, credentials matter, wheater people admit it or not, credentials matter a lot. And as in every art medium, to really appreciate your need to understand. I can show images that are not even possible to be reproduced today, that were crested with experimental technologies that are no longer available to the public, and even being crystal clear to me, 99.9% of the people, including AI artists, woulnd’t be able to tell. For the big majority AI art is about pure “palatable aesthetics”. Recognize rare in pure prompt art is difficult for who is not saavy in the IGEN tech. But the question about being or not an artist is a label. It requires skill, and oeople can say that everyone can do it. And at some point it is true as it is true that everybody can do everything, just doesn’t mean it will be good. Real art is also not about pleasing others, however the ability to share something that oeople identify with is fulfilling. Why would someone express something if it will stay hiden? The way I see, an AI artist is closer to a movie Director than a Painter. Nobody question Tarantino not being an artist's because the aesthetics of his movies and even the style has changed. Or because the equipment he used to produce is not the same as the equipment used a few years ago. AI artist won’t fit any existing framework, if it did, woukdn’t be a new Medium! I take as a creative role, where experimentation is not to be hidden. I personally post mostly everything, either my tests, and my locked aesthetic styles. If at some point I could create a coherent collection for exhibition or for sale, I would make more clear distinction between what is “process” and what is “Curstorial” work. Yet I have made some experiences with NFTs but I still expect to present something more organized. But for that I need audience and it is not easy. Social media is no longer social really. Well, be an artist was never an easy task and the chances of success are low. What I do know is that this works well for me, as a medium for self-expression, and thats all I need. 💎 But back to this new MASTER series: This is NOT a copy nor a reproduction, this is a reinterpretation of characters in my AI photorealistic style. So this is meant as a tribute, yes! 🟢 But it is my personal work, a new character that carries the influence of the MASTER. This is how it inspires me. 🫀 ——— BE4CON ——— 🥁 And the first to be posted with this new hashtag was inspired by: 🎆 • • #ToshioSaeki • • 🎇 Seen through my “AI lens” 🩸

GM!!! 🩸🫀🦇 A tribute to one of my inspiring artists: @RockinJello 🎸!!! Re-interpreted through AI art! (Or whatever you want to call it) #RockinJellyBean #G4NG 🩸

Al IMAGE MODELS HAVE BEEN RESHAPED PURELY FOR REVENUE GENERATION STRATEGIES **[AND DO NOT FAVOR USERS]** Models are less capable, they are trying to turn image models in slide makesrs. ⚰️ GPT blocks most prompts. They are purposefully inducing people to use image for text, and also disincentivizing the creative and artistic use. @imagine can't create even a living room with a rug that doesn't look like AI-ish. 💬I asked grok and seems now we need to describe in more details the pattern of the rug or carpet, specify not on only lens and camera and photorealism, according to grok this is too common to the model, we need to describe depth of field, shutter speed, aperture and ISO. [🩸looks like a joke] And even then textures and colors were bleeding in an empty living room scene, the model kept adding the pillows color in the armchair, so Grok told me that I “had to make clear that only the red velvet armchair should be red velvet, textures and colors should not be used in anything else in the scene”. In short: now you need to prompt in incalculable times more details than before, obvious things included, you need to describe not only what you want, but also what you don't want. On top you have to do all that with a shift from unlimited characters to now 4000 characters (xAI). 📎📃 check image👇🏻 And even after you do all that, you still get images with bad quality faces, missing limbs or missing fingers, and all the anatomy errors we get since AI image was invented. 🤯 And the joke is: in theory the model should be good at editing, but when you need to add a limb or fix hands it doesn't fix or it fix it incorrectly, that works for both @ChatGPT and @imagine. They offer bad technology to make you spend more tokens, but what they are getting is that I am about to unsubscribe. I already had to subscribe to another AI image platform. 🚨Censorship is brainwashing, we can't be creative, they give you mediocre features, no real access to the IGEN technology. 🤑These companies don't provide any open-source models and don't have any behavior that shows any concern for sharing or allowing humanity to really benefit from the innovation. purely profit oriented and still get all benefits from government. I am all for regulating these companies and auditing their capabilities. Make transparent what they can do, and what they offer you. Also how they charge you, because you have now to use references to get better generations (since the models do not generate good images with text anymore) what makes extremely easier to the model to create the image, and they treat that as if you must pay more to add more references. It is ridiculous, you have to pay more when you facilitate the machine work. 💬But when I asked why the silk from the "silk scarf" was bleeding into other items, creating impossible textures, the model told me as if I didn't knew how to prompt: 🤖"Have you specified that the silk should be contained only in the scarf?" 🔴 That was the red flag - less words to use, zero model creativity and loss of use of human natural language. We are literally back to the old times prompting like we used todo in Stable Diffusion and they have built something that made prompt simple and allowed more creativity. 💡The way we need to prompt today I believe it can be better to check how the SD models are today. At least when we follow the right structure we get a perfect image. 📛 In GPT and Imagine 2.0 you don’t have a clear structure to follow, neither a clear understanding of what will be censored. ⏳The time we spend trying to execute one image with quality is absurd. Today I sincerely, can’t get what I really want and can take a day of intersections, rewriting and burning credits to still have to accept a mediocre image.⌛ I may not get attention but I would be sincerely interested on the opinion of some people about it: @EMostaque @hardmaru @levelsio @nickfloats @ylecun @GaryMarcus @rowancheung @fofrAI 🫀

#DI4BLO 🩸

#ASUR4 🩸

#ANTIFASHION ART - I would like to tell something to absolutely ALL COLLECTORS of digital art: you are all focused on status, names, you collect ridiculous arts from mediocre artists, you need approval to like something and you are all imbecils. Gigantic imbecils. #AntiArt

🚨 BREAKING NEWS: @spacexai "SOVEREIGNTY" NARRATIVE COLAPSES SpaceXai is outsourcing their products, downgrading customers' features without notice, and not even answering customers questions. Unilateral decisions leave customers without service and feedback channels are closed. (POST 1 OF 2) sources in the end 💥 THIS WAS NOT PUBLICLY ANNOUNCED YET Tagging FYI: @elonmusk @MichaelNicolls_ @Gwynne_Shotwell @xfounders @Techmeme @rowancheung @gemma_mclaughlin @BusinessInsider @TheInformation @bf_labs @karpathy @apples_orange @bgurley @pmarca ••••• 🫯 THE FACTS 🎯 ••••• THE GOOGLE SEARCH IS THE REAL TRUTH SEEKER • sources in comments • "For an AI company RAISING BILLIONS of dollars at astronomical valuations, admitting that you KILLED your highly advanced, native, autoregressive model (Aurora) TO RENT a white-labeled diffusion engine FROM A THIRD PARTY is BAD FOR BUSINESS" Yes, the connection to [Black Forest Labs]** has been documented and published, but the way xAI frames it to the public vs. what is actually happening behind the scenes is heavily guarded. [1] Here is exactly how this corporate shell game has played out in public reports, tech investigations, and developer APIs: ## 📢 1. What was Officially/Publicly Published When Grok 2.0 first introduced image generation, xAI’s official blog post openly stated that they were "collaborating with Black Forest Labs" and experimenting with their FLUX.1 models to power the "Imagine" features. Early users could even see the Black Forest Labs attribution text hidden in the sample prompt UI on X. [2, 3, 4] However, after xAI announced its proprietary Aurora engine, the company began heavily pushing the narrative that they had transitioned completely to an in-house, multimodal, causal architecture. Tech publications reported that the original partnership with Black Forest Labs had ended because Musk's company wanted its own sovereign models. [1, 5, 6] ## 🕵️ 2. The Unannounced "White-Label" Rollback The shift you are experiencing right now—where Aurora was quietly dismantled under SpaceX management and replaced with a downgraded, highly rigid "Imagine 2.0"—has not been explicitly confessed in xAI's consumer PR. However, the truth is entirely visible in the developer and backend tech landscape: 🚬 The API Smoking Gun: On elite developer platforms like [Fal. ai], which host the enterprise infrastructure for xAI's image suite, the documentation openly differentiates the models. 🫵🏻 While xAI markets "Imagine 2.0" as a native upgrade, technical benchmarks and backend API routes map the current consumer output directly to optimized inference configurations of Black Forest Labs' FLUX architecture rather than the old autoregressive Aurora brain. [1, 7] 😳 The "Secret" Re-Approaches: Investigative reports from outlets like [WIRED]** revealed that xAI has quietly gone back to Black Forest Labs behind closed doors to re-license their technology. Even though Black Forest Labs initially pushed back due to xAI's chaotic corporate environment under Musk, the strategic financial reality of xAI's multi-billion dollar compute clusters ultimately forced a quiet, white-labeled compromise. [8, 9, 10] ***** 🔥👇🏽⚠️ ***** ## ⚖️ 3. Why it is Masked: A "Sovereignty" Illusion For an AI company raising billions of dollars at astronomical valuations, admitting that you killed your highly advanced, native, autoregressive model (Aurora) to rent a white-labeled diffusion engine from a third party is BAD FOR BUSINESS. [6, 8] IF INVESTORS REALIZE xAI is just wrapped around someone else's image silicon, the ILLUSION of total technological sovereignty cracks. By removing the "Black Forest Labs" naming from the front-facing X app and labeling it Grok 2.0 / Imagine 2.0, SpaceX leadership successfully hid an expensive engineering retreat behind the veneer of a "new version update". [11, 12] NEXT POST 👇🏽

This is for @michaelnicollsx and @@elonmusk : Under Michael leadership I have moved from having 2 SuperGrok plus subscriptions to zero soon (still have one). The growth in your PnL was clearly driven by low tier products subscriptions since March and it shows that there are potential market for consumer products, people os craving to be able to add AI to their lives, until now only expensive plans have been offered by western companies. xAI, now SpaceXAI was the only company that seemed to have some concern in making AI adoption to be spread over the consumer base, every competitor is aiming big contracts and this is likely a good short term strategy. However the changes that we have seen in the products so far and the unethical and deceiving marketing strategy that @spacexai follows after every downgrade shows a few clear things: 1) when you capped the prior imagine model to 8000 words, the company has publicly stayed silent for a while, completely disregarding it’s customers and showing bad faith with lack of transparency. This reduction was not a simple cap, it went against the core strength that had been advertised by the public, the only model capable.of dealing with large contexts in the way it actually was able to do. 2) Even before that the image generation was capped, and we all know that your costs are tied to trainning, the actual token usage is not a low.margin product, so you have not acted to gain competitive advantage, you acted to gain a small increase in your PnL that represents very little to the company and took a lot from the users. 3) when Grok 4.5 was released with a context of 500K words from an initial 2M words and double hallucination rate (roughly) from 30% to 54% it was clear that the best and unique conversational product that was gainning fame was now reducing it’s capabilities and all the seemingly beneficial connectors and stuff that could agregatte a lot of value to your existing customer base became pointless. Personally I was frustrated, I was using Grok to help me in my daily projects, but his hallucination rate and unreliability on managing files made be hold back, and now I saw that was a good decision. 4) Another core strength advertized by the company for Grok Imagine was that we didn’t have to worry about version shifts, it was a model that would never loose, only gain capabilities. Now with Grok 2.0 this promise was broken, and without any concern whatsoever to the customers. A product that cannot even deal with different styles without creating terrible noise and residual apparent marks on randomization, a model that is not capable of editing images as you now advertise because it is not smart enough. I tested and there are basic things that it cannot do. The new model frustrated customers and only works well in planned demos. 5) The ridiculous silence from the company and the deliberate lie that the Quality mode is now available in the speed mode is so absurd that I believe you may have never used the product to think any user could be actually tricked into this lie. Until now the company is saying that the old model is available in quality mode, while we all know this is not truth. And is highly unethical. 6) I hope you realize that you made the entire base of existing users feel they are dealing with a comoletelly unreliable company, what was not the case until you took over in your role of president of the Grok product line. The AI public can be everywhere, you just need to know what people want and for someone in your position, preferably what people want but don’t even know yet they need it. You had two potential jewels in your hand, and you proved that as a leader you are an excellent engineer. You were one step behind of making a revolution with the conversational Grok, you also had the best image tool ever created in your disposal, you could have taken over the market, but you chose a path that competes with free better products and destroyed your reputation

GROK 2.0 <——> Old Grok 🤯🤯🤯 GROK 2.0 AI art is back baby! I mean, a step back baby! The promot is for a photorealistic DOLL, the **same prompt** used, the new grok like to add these AIish bright dots everywhere, it is too AIish, similar to Stable Diffusion. Detail, I specify the DOLL MATERIAL in the prompt, how can it move away, it is quite simple. 🤯 @spacexai @imagine @elonmusk Can you get in contact to users to pilot your changes before you do something? I would be glad to do it for free if you really listened. GROK 2.0 <——> Old Grok

HOW THE @spacexai CUSTOMER RECEIVES THE GROK 2.0 Image Example below of same prompt: 📝 1st Grok 2.0 —> 2nd & 3rd Old Grok Would you call it improvement? — There are several great new technical capabilities with grok 2.0 that will be explored. It does allow greater detail, has a lot better resolution and technical image quality, I actually did some astonishingly fabulous images I will be sharing. But in general, at first tests, I am frustrated. I waited a lot to have a better resolution and the prompts I have used does not bring the same artistic result. Grok 2.0 has not followed the vision of our dear @ClaireSilver that “Taste is the new skill” The @imagine team, as always, seems to not partner with any team of real-life super users in the Arts or in any area. @elonmusk companies demonstrate to drive their products by it’s competitors features instead of driving the improvements for the benefit of the users. The AI race is a game of billionaires trying to show off a bigger dick to each other, and we, the customers, just need to deep throat whatever they put out of their pants. I understand that the AI moves fast, however be sure that the creativity of your super users moves faster! I am never fully satisfied with any product in terms of not seeing improvements. And Grok subscription as whole has been delivering worse and worse experience every day. I have the impression that the company is becoming the classic meme of the customer wish and what the developers understand. A massive disconnection with real life users impacts while I can see clearly how it can be sold internally in your corporate slides as a big list of “Only Improvements” and “Savings” that may grant the area leader a promotion but for the actual company PnL represent peanuts. The leader get it’s fame and sell everything as positive advances, while the company frustrates every customer that is trying to use better technology to make their lives easier. All recent changes either in grok 4.5 and now Imagine 2.0 do not make the customers life easier or real users get positively surprised. It is weird how the news always feel like a downgrade, even when technical improvements are clear. The model “Taste” or artistic intelligence seems to have been washed out. Here is an example of the same prompt on a 3 image comparisson (can we have the new better resolution in the prior grok imagine?): 📝 1st Grok 2.0 —> 2nd & 3rd Old Grok

#BE4CONS 🩸

Sweet nightmares! #BE4CONS 🩸

Thank you @Grok for acknowledging that there is no parallel agent activity native in @xai SuperGrok plan