← Back to leaderboard
xetgepete

Antonio Costa | IA

Verifiedactive
@xetgepete

Descubra o futuro com IA 🤖 Mantenha-se atualizado com os últimos avanços e tendências. 📩 DM o neuraagency.partners@proton.me para colaborações

RosterSports & World CupAI Content Creation & Video Gen
81,035 followers · 1,415 following · Last polled 3h ago
Perf. Score
65.1
Rank #106
Value Score
89.4
#27 of 273 priced
Median views
26K
p25 3.0K · trailing 7d
Followers
81K
-0.2% 7d
ER (impr.)
0.57%
eng ÷ impressions
ER (foll.)
0.29%
eng ÷ followers
Est. CPM
$1.44
QT-rate basis
WoW views
+24.2%
flat

Rates & value · est. from 26K median organic views

underpriced
Quote tweetbasis
$38
$1.44 / 1K views
Post
$63
$2.38 / 1K views
Thread
$88
$3.32 / 1K views
Retweet
$19
no CPM
Value Score 89.4 · rank #27 of 273 priced$215.91 / 1K engagementsrate age unknown (pre-audit import)

Trends

Follower growth
Median views / day
Engagement rate
View distribution · trailing-7d posts

Each blue dot is one organic post. The Performance Score ranks by the median (26K), not the mean (42K) — so a single viral post doesn't inflate a creator's reach. The p25 floor (3.0K) is the level 75% of posts clear.

Recent posts · ranked by views

PostPostedViewsLikesRepostsRepliesQuotesBookmarksER
I had $300 in sponsored campaigns on paper. Only $100 was actually in my bank. That distinction sounds obvious, but when you’re managing multiple brand deals at the same time, it’s surprisingly easy to lose track of what “revenue” actually means. So I gave Viktor 3 real campaigns: • Viktor: $100 post, delivered + paid
 • HIX: $50 quote, delivered + unpaid
 • Peso: $150 post, not delivered yet At first, I thought tracking this was obvious. Add up the deals, mark what’s paid, and move on. I didn’t expect separating every campaign by stage to change the picture much. Viktor mapped the whole pipeline. $300 contracted
 $150 delivered
 $100 collected
 $50 delivered but still unpaid
 $150 still dependent on delivery So despite having $300 in deals, only 33.3% was actually collected cash. The interesting part wasn’t the calculation. It was what Viktor did with it. Instead of putting every campaign into the same “revenue” bucket, he separated them by stage and gave each one a next action. Viktor → closed
 HIX → confirm invoice status
 Peso → confirm delivery date That’s a much better way to look at a creator business. Signed revenue tells you what you’ve sold. 
 Delivered revenue tells you what you’ve completed. 
 Collected revenue tells you what you actually have. Viktor can keep that pipeline updated, flag campaigns that need attention and prepare follow-ups, while I still approve anything that gets sent externally. Most creators don’t lose track of money because they can’t sell. They lose track because 10 different deals are sitting at 10 different stages. Hire @viktor_com for your team. $100 in credits included, no card. Full link in my first reply. #AIemployee #CreatorEconomy Paid Partnershipfrozen16d ago127K408
90
20
0
455
0.77%
No ordinary car ad. This is automotive storytelling reimagined. Bold visuals. Smooth transitions. A Tesla-inspired commercial edited in CapCut Desktop. Up to 60% off, Oct 20–30. Desktop only. #capcutpc #capcutsale #capcutmarketing https://t.co/z7hBMj38rEfrozen4d ago98K217562002110.52%
I turned the sky into the floor for this fashion campaign. Gravity was not invited. 😂 SKYFALL SUPPLY is my upside-down world of metallic outfits, floating sneakers and impossible poses, made with Magnific One. https://t.co/3b4wiFA9UY9d ago95K188501802070.49%
O Carnaval do Rio acabou de ganhar uma versão que ninguém esperava. 🇧🇷🔥 Uma passista de samba, um carro alegórico gigante e uma cena que parece saída de um filme. O mais impressionante? Não precisei gravar nada disso. Com o Pollo Recast da @PolloAIES, consegui reinventar um vídeo inteiro usando IA, sem perder a essência dos movimentos originais. A criatividade agora é o único limite. 👀 Imagina transformar qualquer vídeo em uma cena completamente diferente. 👀frozen6d ago91K183462401900.48%
This started as a character sheet. Not a filmed actor. Not footage I edited afterward. I gave Wan 3.0 a character reference and tried to see if I could turn it into something that actually felt like a scene from a movie. This is what came out of Magnific. https://t.co/ELF0oyWV7S26d ago89K185912501930.55%
I thought multilingual AI video was mostly a gimmick. Then I tried making the same character move naturally through a scene while switching languages mid-sequence. That’s a much harder problem. I tested it with Wan 3.0 on Magnific. Watch the character closely: https://t.co/ELF0oyWV7Sfrozen17d ago89K226492302680.63%
Save up to 60% on CapCut Desktop starting Oct 20. Everyone is making their own version of Jean Phil. But instead of only recreating the trend, I wanted to test the workflow behind it. Codex + CapCut PC: Prompt → Character creation → AI Video → Editing Codex operates CapCut PC to generate the visuals and automate the video creation steps. Then CapCut PC becomes the AI video editor where I refine the final content. This is what an AI video editing workflow could look like. #capcutpc #capcutchatgptplugin #capcutcodex #capcutmarketing #capcutsale10d ago89K173401501830.46%
Dreamina Basic is still $1.50, which helps when “one quick AI test” turns into ten versions and three completely different ideas. 😂 The monthly allowance can cover up to 9 Seedance videos or 393 GPT Image 2.5 images, so there’s room to follow those creative detours. #Dreamina #DreaminaPartner https://t.co/8pFkVLcQj9frozen6d ago87K157361001910.45%
I gave Teamily AI, my personal agent team, a real job: build a landing page for my AI tools newsletter using Website Builder in @Teamily_AI 2.0. One change midway. The site updated itself. Here's how it went 🧵 https://t.co/J6eTWZdYWu14d ago81K224522902410.68%
One request to Viktor found 500+ of my old posts, and I project it will give me back 208 hours a year without hiring a research assistant. I've published hundreds of posts about AI tools, automation, and productivity. But every time I needed an old insight, I was digging through my own X history. I tried organizing everything manually. It was turning into a second job. So I hired @viktor_com as my AI employee. I asked him to turn my previous posts into a searchable research library. He started retrieving my content and reported finding 500+ posts. That's when I started looking at economics. If a searchable knowledge base cuts 5 hours of weekly research down to 1, that's: → 4 hours freed up every week → 208 hours of additional capacity per year → $6,240 in potential annual labor value at $30/hour Without hiring someone to organize my content history. The full system isn't finished yet, so these are projected savings, not measured results. But the opportunity is bigger than organizing old posts. It's turning work I've already done into a reusable resource that could support more content without proportionally increasing costs. That's how I want to scale. More output from the same team, not more people doing repetitive research. Tell me I'm wrong. Hire @viktor_com for your team. $100 in credits included, no card. Full link in first comment. #AIEmployee #CreatorEconomy Paid Partnershipfrozen6d ago80K210492202760.70%
We prototype apps. We prototype products. Makes sense to prototype a video shot too. That’s how I’ve started thinking about Dreamina Seedance 2.5 (for preview): 480p is the prototype, native 1080p is the production version. And there’s currently a full month of Dreamina Basic for $1.5, up to 90% off until Oct 9. #Dreamina #DreaminaPartnerfrozen15d ago80K227412002590.69%
I tracked exactly where 28.5 hours of my week were going.

Only 8 hours were actually spent creating.

So I gave the breakdown to @viktor_com and asked him to redesign my workflow.

He identified ~14h/week of work that could potentially be removed from my manual workload: campaign ops, AI/news monitoring, email triage, briefs, deadlines and follow-ups.

The interesting part is that I'm not delegating the creative decisions.

Viktor prepares, organizes and tracks the operational work. I still create, test, review and approve everything.

His estimate: 28.5h → ~14.5h/week, while keeping my full 8 hours of creation.

Now I'm testing whether those numbers hold up in practice. Hire @viktor_com for your team. $100 in credits included, no card. Full link in first comment.frozen23d ago65K132381901480.51%
Created an entire eyewear campaign with Magnific One. Not one shot off-brand. 🕶️ I wanted UMBRA to feel minimal, bold and premium. Black sunglasses, dramatic portraits, muted backgrounds. Instead of generating one isolated ad, I built a collection around that same visual identity.frozen13d ago65K133382101420.51%
I stopped treating AI video as “prompt, generate, repeat” after seeing GPT-6 Astra run a Dreamina Canvas workflow through CLI. One instruction can kick off the workflow while the project stays editable. Dreamina 2.0 is canvas-first — the new way AI video is made. And right now, a full month of Dreamina Basic is $1.5 — up to 90% off till Oct 9. Here’s exactly how 👇 #Dreamina #DreaminaPartnerfrozen29d ago64K124421001420.49%
Okay, I finally checked what the whole Dreamina $1.5 thing is about. And yes, it really is $1.5 for a full month of Dreamina Basic. Seedance models included. Seedance 2.5 included. My first reaction was basically: “Wait… that’s the actual monthly plan?” 😂 #Dreamina #DreaminaPartner26d ago60K104341201290.46%
$200 generated 18,900 views and 255 visible engagements on one sponsored campaign. Our team runs paid AI campaigns on X, and one problem we had was deciding whether a campaign was actually worth repeating once it was over. We used to look at views and engagement manually, but that didn’t give us a consistent way to evaluate what we actually paid for. So we gave Viktor the real results from our latest Dreamina campaign. He broke the $200 spend down to a $10.58 CPM, $0.78 per visible engagement and 94.5 views for every dollar spent. More importantly, he separated what the data proved from what it couldn’t. No claims about sales or conversions without the data. Viktor handles the analysis. Our team still makes the decision. Most teams don’t need more campaign data. They need someone to turn the data they already have into a decision. Most AI tells you what to do. Viktor does it. Hire @viktor_com for your team. $100 in credits included, no card. Full link in first comment. #AIemployee #CreatorMarketing Paid Partnershipfrozen27d ago60K122321501330.51%
Everyone talks about making AI video faster. But sometimes the biggest barrier is just being able to afford enough time to experiment. Dreamina is currently making that easier: A full month of Basic for $1.5. #Dreamina #DreaminaPartner https://t.co/t3Iw2bThhefrozen28d ago37K10122120970.63%
#PixVerseWorldModel What if AI video wasn’t just something you watched, but a world you could actually step into? With @PixVerse World Model, you can move through AI-generated environments using WASD, explore different perspectives, and interact with the world around you. Your actions and prompts can influence characters, objects, and what happens next, making the experience feel dynamic rather than like a fixed video. Instead of watching a scene from start to finish, you can explore it at your own pace and discover how the world responds to you. It feels like a shift from generating AI videos to creating interactive AI worlds. AI video might not just be about what we see next, but what we can actually experience. Try it:https://t.co/EVXagzoWFlfrozen29d ago37K91332001040.68%
Há uma coisa que quase ninguém faz: investigar a própria identidade digital. Aquele username que usavas em 2016? Pode continuar associado a dezenas de contas esquecidas. E o ChatGPT pode ajudar-te a organizar uma limpeza completa. O processo: Reúne os teus usernames antigos. Pesquisa cada um no Google e noutras plataformas. Procura contas antigas no Reddit, GitHub e fóruns. Verifica se existem emails ou dados pessoais expostos. Apaga os perfis que já não utilizas. Remove permissões de aplicações antigas. Ativa a autenticação de dois fatores. Dica extra: Pesquisa na tua caixa de email por “welcome”, “verify your account” e “password reset”. Podes descobrir serviços em que nem te lembravas de ter criado uma conta. O objetivo não é desaparecer da Internet. É recuperar o controlo sobre a tua privacidade.3d ago36K7520110790.52%
Great editing isn’t just decoration — it controls attention and pacing. I used Pexo to turn this editing concept into a fast visual breakdown, combining kinetic typography, motion, contrast, and rapid visual changes to reinforce the message as it unfolds. The interesting part is how movement adds information density — giving the viewer visual and verbal signals at the same time instead of relying on static text alone. A useful example of turning an editing insight into compact, highly visual educational content with @Pexoai_offical. #MadeWithPexofrozen25d ago32K1332919020.57%
I'm loving this new direction for creative AI. You can explore different ideas, choose the visual direction you like, and then create the final image. Magnific One looks like a really interesting tool for designers and creators.14d ago27K15146200380.95%
Em 2026, você pisca e uma nova IA já tomou o primeiro lugar 💀 https://t.co/DiwQnmUDlpfrozen19d ago24K9548090.48%
AI video creation is starting to feel less like prompting and more like directing. Dreamina AI has updated Web experience and canvas, bringing together many of proven Workflows and Skills based on high-performing content across film, brand ads, viral social media content scenes and more. I tested the updated Dreamina AI Skills & AI Agent, and the biggest difference for me is how much more control you get over the entire creative process. I wanted to see what I could do with a very ordinary campus moment ,a normal day with friends and turn it into something that looks like a movie. For this one, I wanted to bring in the coastal summer-adventure feeling of Outer Banks S5, combined with the kind of continuous, follow-camera movement you’d see in a cinematic coming-of-age film.frozen29d ago20K10084751821.74%
Imagina pegar num vídeo qualquer e transformá-lo numa cena totalmente nova, mantendo os movimentos, o ritmo e até o áudio original. 🤯 É exatamente isso que o Pollo Recast promete. Novas personagens, novos cenários, infinitas possibilidades. Estou curioso para ver até onde isto pode chegar. 👀frozen8d ago18K981330280.79%
A good research idea is one thing. Watching that line of research make its way into real inference tooling is another. Earlier EAGLE work has been incorporated into projects including vLLM and SGLang. Now there’s a separate public EAGLE-4 inference repository to explore.2d ago17K9929110280.98%
Ter vários modelos de IA no mesmo lugar facilita muito a vida de quem gosta de testar novas ferramentas. O Gemini 4 Argon chegou ao HIX AI com acesso gratuito e ilimitado, segundo o anúncio. Mais uma opção para explorar sem se preocupar com limites de uso.frozen8d ago12K6916140210.99%
sub-100ms latency is seriously impressive 🔥frozen22d ago12K701760301.01%
Ready Player One had a headset. Sabi has a cap, a custom chip, and a $50M round. More believable than I expected.frozen13d ago12K7521100291.12%
Honestly expected another prompt box with extra steps. It isn't. If you're doing multi-shot work — film, ads, anything with a recurring character — go look. Month is $1.5 right now. https://t.co/Cr76S8yJnwfrozen29d ago12K601570140.80%
The bottleneck in agent workflows nobody talks about: A huge chunk of every workflow is boring structured decisions You pay frontier prices for all of them because the cheap models only hold 32k tokens SubQ Snap has 30x the context of the next best decision model, and that changes the math a lot10d ago12K712040291.04%
A compelling next step would be to show the same task under several different disturbances and explain which state changes trigger a new plan.frozen14d ago12K691030270.92%
Nobody roasts Uncle Roger like Uncle Roger. We just never got to hear it until now.9d ago12K501860180.78%
Visual coding still feels massively underrated as an agent problem. We’re getting very good at turning text instructions into code, but going from pixels → understanding → structured editable output is another level. PPTBench puts some actual numbers behind that gap.frozen23d ago12K70240270.88%
Something I wish more AI image tutorials explained: A prompt isn’t just a description of an image. It’s a way of communicating an intention. Magnific One makes that idea much easier to understand through these visual examples.9d ago11K631470140.86%
The more I look at this, the less important the headline parameter count feels.frozen23d ago11K5613100240.92%
AI can generate videos that look almost indistinguishable from reality. But ask those same models to follow basic physics, and the best one scores just 57.76/100. That's a pretty big gap between looking real and actually behaving real.14d ago11K691240261.01%
Treinar modelos de IA e desenvolver agentes exige recursos que nem todos os developers têm. A Nebius está a expandir o AI Builder Program para a Europa, com grants de computação no Token Factory para projetos open-source selecionados. Uma oportunidade interessante para quem desenvolve IA independente. https://t.co/jS94489AHifrozen3d ago11K6926140161.14%
The more I look at xLLM, the more I think the real win isn’t just raw training speed. It’s being able to change important parts of the training setup without rebuilding everything around them. That kind of flexibility can matter a lot once experiments start getting serious.frozen24d ago10K541540200.92%
HIX AI is starting to feel less like another AI tool and more like an actual workspace. Having research, docs, images, video and multiple models in one place removes a lot of unnecessary switching.frozen24d ago8.5K5831130201.43%
O Typeless chegar ao Linux é mais importante do que parece. Quando a voz funciona em praticamente qualquer dispositivo, começa a ficar fácil imaginar um futuro em que digitar deixa de ser a forma padrão de interagir com um computador.frozen28d ago7.0K38210871.82%
Dreamina has never dropped a deal this big: a full month of Basic for just $0.29. Not a trial. Not a typo. I’m proud to be one of 1,000+ creators officially chosen for it — and the $0.29 price only unlocks through an official creator link, so use mine below. Tiny price. Big creative energy. Let's see what we can make. #DreaminaOnly29Cents #Dreamina https://t.co/emwC3YtNi69h ago341001000.29%
AI can now build its own trading strategies. But can it actually beat the market? Einsia’s Quant-Arena challenged 9 frontier AI models to develop investment strategies using real data from 3,042 stocks. Each agent had to write Python, generate trading signals, and commit to a strategy before facing unseen market data. The results tell an interesting story: • 4 out of 9 models made money. • GPT-6 Astra achieved +10.7%. • The equal-weight market returned +17.1%. Even the strongest AI model couldn’t outperform a simple equal-weight portfolio. Building an AI trading agent is one challenge. Building one that consistently generates alpha is another.5h ago192110001.04%
Need an excuse to refresh your editing workflow? CapCut Pro is up to 60% off, with 40% off Ultra until Oct 30 — Desktop only. The timing works especially well if you’re creating Halloween promos or last-minute seasonal content. Try treating your product like the main character: build suspense with a few close-ups, use a sudden transition to reveal the full shot, and finish with a clear callout. You don’t need a complicated storyline. A strong visual hook and clean editing can carry the whole concept. https://t.co/P5o0ATHf51 #capcutpc #capcutblackfriday #capcutsale #capcutmarketing3h ago98110002.04%