TechwithTeddy

Helping beginners use Tech & AI to grow their skills, business & income 🚀
Street Tech Quiz 🎤 | AI Tools 🤖 | Digital Skills 💻 Founder @techwithteddyin

[TechwithTeddy @techwithteddy]

I stopped burning AI credits trying to discover the right camera movement.
I directed the entire shot in @blender first, then used that previs to control the generation.
Final result: 10-second AXIOM shot for ~$1.15 of GPU generation compute on @runpod using self-hosted MiniMax H3. @MiniMax_AI
No monthly AI video SaaS subscription.
No burning credits on endless camera retries.
Blender = direction. AI = generation.
I’m building the whole workflow in public.
Want the playbook?
Reply “SELFHOST” and DM me.

[TechwithTeddy @techwithteddy]

@JagTheHero This is exactly the economics we’re experimenting with at TechWithTeddy - MiniMax H3 + ComfyUI + RunPod/cloud GPUs instead of constantly buying generation credits.

I’m documenting the workflows, costs and tests as I build them and teaching the setup FREE in the community.

Anyone trying to build their own self-hosted AI video stack can join here:
https://t.co/m5TvLqBoEP⁠�
#TechWithTeddy #MiniMaxH3 #RunPod #ComfyUI

[TechwithTeddy @techwithteddy]

This is exactly why I built the TechWithTeddy community.
AI video becomes VERY expensive when every failed generation means buying more credits.

I’m currently building and testing a different workflow:
MiniMax H3 + ComfyUI + self-hosted models + RunPod/cloud GPUs + Blender for control.
The goal is simple: generate more, waste less, and stop depending entirely on expensive AI video subscriptions.

And I’m teaching the entire setup FREE — from getting the models running to building practical AI video workflows.

If you’re experimenting with MiniMax H3, self-hosted AI video or cloud GPUs, this community is being built specifically for you:
https://t.co/m5TvLqBoEP⁠�
#TechWithTeddy #MiniMaxH3 #AIVideo #ComfyUI

[TechwithTeddy @techwithteddy]

@RechterRoni @juliarturc This really cracked me up. 😂

[TechwithTeddy @techwithteddy]

@rirahcreates Lol

[TechwithTeddy @techwithteddy]

This is the real question to ask.

If Codex is editing a 1-hour YouTube video, how much are we actually spending on tokens, API calls and compute? At some point, the automation cost could approach what you’d pay an editor.

I’m experimenting with this kind of AI video + YouTube automation workflow at TechWithTeddy, so cost per finished video is one metric I really want to benchmark.
@BrightAfeno have you calculated your actual cost per 1-hour video yet?

#TechWithTeddy #YouTubeAutomation #Codex #AIAutomation

[Bright Afeno | Fenoxx @BrightAfeno]

Faceless YouTube, my arssss…
I’m done!

I just started learning YouTube automation…
Learning, testing , and building stuff inside IDEs almost at zero cost.

Let’s see how YouTube automation goes…

[TechwithTeddy @techwithteddy]

New device alert: Starlink added to the TechWithTeddy workflow.

One thing I’ve been working on is removing infrastructure limitations from what I can build.

With AI video generation, self-hosted models, cloud GPUs, Blender, Codex and automation, a reliable high-speed connection is becoming just as important as the computer itself.

So, Starlink is officially part of the setup.

Highest speed I’ve seen so far: 300 Mbps.

More bandwidth. More cloud computing. More building. More experiments to share.

The TechWithTeddy lab keeps growing.

#TechWithTeddy #Starlink #CloudComputing #AI

[TechwithTeddy @techwithteddy]

@Raynerdtech @Cloudflare Why is @github not first on the list? 😭

[TechwithTeddy @techwithteddy]

@Ekitipikin this is impressive for your second AI video.Quick question: which AI video platform did you use, and roughly how much did it cost you to generate this 36-second video, including the retries?

I’m currently testing self-hosted AI video + ComfyUI + cloud GPUs at TechWithTeddy, so I’d love to compare the production cost.#TechWithTeddy #AIVideo #GenerativeAI #ComfyUIThe “including the retries” part is important.

A 36-second final video might contain only 36 seconds of usable footage, but the real AI-generation cost can be much higher if several clips were regenerated before getting the final shots. https://t.co/duHOrii6GK

[TechwithTeddy @techwithteddy]

This is @Ekitipikin’s second AI video, and it’s already doing serious numbers.

Now I’m curious about the production side:

How much did this 36-second AI video actually cost to generate, and which AI video platform did you use?

This is one of the problems we’re working on at TechWithTeddy using self-hosted AI video models, ComfyUI and rented cloud GPUs to reduce dependence on subscriptions and expensive generation credits.

I’d genuinely like to compare the cost of this workflow with what we’re building.

#TechWithTeddy #AIVideo #ComfyUI #GenerativeAI

[TechwithTeddy @techwithteddy]

This is @Ekitipikin’s second AI video, and it’s already doing serious numbers.

Now I’m curious about the production side:

How much did this 36-second AI video actually cost to generate, and which AI video platform did you use?

This is one of the problems we’re working on at TechWithTeddy using self-hosted AI video models, ComfyUI and rented cloud GPUs to reduce dependence on subscriptions and expensive generation credits.

I’d genuinely like to compare the cost of this workflow with what we’re building.

#TechWithTeddy #AIVideo #ComfyUI #GenerativeAI

[Mayor Of Ekiti @Ekitipikin]

Please rate my second AI video 😌😌 fast learner 🤪

[TechwithTeddy @techwithteddy]

@Aje_Dynamicz Since you’ve already verified your account and tried Facebook Lite, I have another idea.

I can create the Facebook Page from my end and add you as an admin with full access. No charge, and I won’t need your password or login details.

If Meta allows me to add your account successfully, that would at least show that your account can receive and manage Page access. The problem with creating a Page directly may then just be an account-specific or algorithm/system inconsistency rather than a general restriction.

Send me a DM if you’re interested and let’s test it.

#TechWithTeddy

[TechwithTeddy @techwithteddy]

@Aje_Dynamicz I decided to test this myself because I’ve seen several Nigerians saying they can no longer create a Facebook Page.

I just created one successfully in less than a minute, as shown in this video.

So this does not appear to be a blanket situation where Meta has stopped every Nigerian account from creating Facebook Pages.

If yours isn’t working, I would first make sure your Facebook account is properly verified and has no restrictions. You can also download Facebook Lite and try creating the Page from there.

Sometimes when a feature isn’t working on one account or app, it doesn’t necessarily mean the feature has been removed for everyone.

#TechWithTeddy #Facebook #Meta #SocialMedia

[Ajé | the Gohighlevel Guy @Aje_Dynamicz]

Pls has anyone created a Facebook Page of recent ?

Been on it for over a week now and it's not working
pls help

[TechwithTeddy @techwithteddy]

5 million ChatGPT Sites in three months says a lot about where software creation is heading.

We’re exploring the same idea at TechWithTeddy with AI coding, Codex and automation: how much of a real product can one person design, build and operate with AI doing the heavy lifting?

Building is becoming accessible to a much larger group of people.

#TechWithTeddy #ChatGPTSites #AICoding #VibeCoding

[TechwithTeddy @techwithteddy]

5 million websites created with ChatGPT Sites in just three months is the number that stands out to me.

Building a website used to mean learning code, configuring hosting, connecting databases and spending days moving between different tools.

Now AI is compressing more of that workflow into a conversation.

At TechWithTeddy, this is the bigger shift I’m interested in: combining AI coding, Codex, automation and generative media so one person can move from an idea to a functioning digital product much faster.

The barrier to building is collapsing. The advantage will increasingly be what you choose to build and how well you execute it.

#TechWithTeddy #ChatGPTSites #AICoding #BuildInPublic

[ChatGPT @ChatGPT]

Three months ago, we launched ChatGPT Sites – an easy way for anyone to build and host fully functional, interactive web apps. Since then, people have created over 5M sites.

We’ve been listening to your feedback and ICYMI, we’ve launched a few Sites updates:

🫂 Build together: Invite teammates to edit, save, and publish to your shared Site

🔐 Share privately: Invite specific people to a private Site without making it public

🚀 Launch faster: Go from prompt to deployment in half the time

👀 Explore your data: Ask ChatGPT to inspect your Site’s database. Editors can view it, too

🌐 Add your own custom domain

[TechwithTeddy @techwithteddy]

The scale alleged here is what stands out.

As Claude, DeepSeek, Moonshot and other AI models become more capable, questions around distillation, training data, privacy and model provenance are going to matter much more.

It’s something worth paying attention to as we explore more self-hosted and open AI at TechWithTeddy.

#TechWithTeddy #ClaudeAI #AISecurity #OpenSourceAI

[TechwithTeddy @techwithteddy]

This is an important side of the AI race that builders shouldn’t ignore.

If Anthropic’s allegations are accurate, using fake accounts to extract large volumes of Claude interactions to improve competing models raises serious questions around AI model distillation, data ownership, privacy and security.

As we build more with open-source and self-hosted AI at TechWithTeddy, I think understanding where models and training data come from will become just as important as knowing how to deploy them.

The AI race isn’t only about who builds the smartest model. It’s also about how that intelligence was obtained.

#TechWithTeddy #ClaudeAI #AIModels #AISecurity

[International Cyber Digest @IntCyberDigest]

‼️ BREAKING: Chinese labs DeepSeek and Moonshot were quietly relaying customer prompts to Claude through fraudulent accounts. Users thought they were using Chinese AI, but were actually getting answers from Claude.

Their customers' sensitive data now is in Anthropic's hands. One user, who is tied to China's military, asked Kimi whether a person tracked across hundreds of CCTV cameras in Chengdu was behaving abnormally. DeepSeek passed along live credentials for a Russian government database, shared by an IT operator handling data from a Defense Ministry-linked agency.

The two Chinese labs relayed customer prompts to Claude through fraudulent accounts and saved the exchanges to train their own models. Moonshot forwarded almost 300,000 customer requests in a single ten-day stretch. DeepSeek flagged users running its models inside coding tools like Claude Code and OpenCode, then routed some of them to Opus.

[TechwithTeddy @techwithteddy]

I wouldn’t throw away those 14 years of editing experience yet.

GPT-6 Astra handling clips, music and subtitles is impressive, but an experienced editor who learns to direct these AI tools could become dramatically more productive.

That’s something I’m actively exploring at TechWithTeddy: how far we can automate video editing with AI + Codex while keeping the creator in control.

#TechWithTeddy #GPT6Astra #VideoEditing #AIEditing

[TechwithTeddy @techwithteddy]

GPT-6 Astra editing 84 video clips, finding music and generating subtitles is a glimpse of where video production is heading.

But I don’t think this means 14 years of editing experience is “done.”

The opportunity is combining that experience with AI.

This is exactly the direction I’m building toward at TechWithTeddy: AI + After Effects/DaVinci Resolve + Blender + Codex + automation, with AI handling more of the repetitive production work while the creator controls the final result.

The editor isn’t disappearing. The editing workflow is being rebuilt.

#TechWithTeddy #GPT6Astra #VideoEditing #AIAutomation

[el.cine @EHuanglu]

never thought AI could do video editing

now i think my 14 yrs editing experience is done

GPT-6 Astra watched 84 clips, edit, trim, find music in my library and even generated subtitles on davinci resolve all by itself.. this is so scary

[TechwithTeddy @techwithteddy]

GPT-Live-1 + tool calling is the part I’m watching.

We’re exploring similar workflows at TechWithTeddy for AI receptionists and business automation. Once a voice agent can naturally listen, respond and use tools during the conversation, it becomes much more than a voice chatbot.

#TechWithTeddy #GPTLive1 #VoiceAI #AIAgents

[TechwithTeddy @techwithteddy]

GPT-Live-1 coming to the API is bigger than simply making AI sound more natural.

Full-duplex voice + tool calling means developers can build AI agents that listen, respond naturally and take actions while the conversation is happening.

This is directly relevant to what we’re building at TechWithTeddy around AI receptionists, lead qualification, customer support and business automation.

The next generation of business AI won’t just be a chatbot you type to. It will be an agent you can actually talk to while it gets work done.

#TechWithTeddy #GPTLive1 #VoiceAI #AIAgents

[OpenAI Developers @OpenAIDevs]

GPT-Live-1 is now available in the API.

Bring ChatGPT’s natural back-and-forth to your app, with voice agents that listen while they speak and work with the models and harness you choose.