Bilal

Can someone check on on the guy who laid 69 mil for that nft? Remember watching a doc about how all his assets were in the metaverse

@ChaseWisely @sweatystartup No you just have to flow the breakthroughs. It's cheaper for Openai to serve gpt 5.4 than gpt 4. These breakthroughs will continue both on the algo and hardware level Demand will rise yes but the pace of optimization will dwarf it. That's what I believe anyway.

@deepwhitman @sweatystartup the one part I think is true is the prices won’t last forever, they are somewhat subsidizing our plan usage so that we become customers and test things etc, it does make sense that it can’t last forever at this pricing or at least not without ads perhaps?

AI about to get 20x expensive. These $200 / month claude subscriptions are burning $5,000 worth of credits. The bubble is going to pop and it will pop soon.

@AIWarper nice thanks!

@deepwhitman Video. Color grading / grain etc. Even automating it would make it better than nothing

Ltx 2.3 audio + image Audio generate with Sonauto ai Not bad tbh... song is a banger, visuals can be lacking at a distance. Desperately needs some upscale / second pass. Will keep fiddling in my spare time

@l2k still feel a regression in model iq when I try using it from codex/claude. What tasks do you feel it does as well/better on?

All the people saying Cursor is dead clearly have not used the product recently - it’s gotten incredibly good

@AIWarper thanks! when you say post grading were you thinking of any particular techniques for audio or video post processing?

@deepwhitman Nice! Some post grading on top of this would push it much further

@G48ST4R @sweatystartup enjoy taking tech advice from a laundromat owner.

@deepwhitman @sweatystartup Denial is the first stage of the five stages of grief.

Congrats. This might be the right time to rethink project-based pricing and move closer to true usage-based pricing. As more users launch multiple projects, the current model adds friction to experimentation. Removing that friction could easily drive higher overall usage and, in turn, more revenue.

There are now 7M developers using @supabase Signups have been accelerating since the start of the year Our growth rate right now is as fast as it was during YC, except that we are doing it from a base of millions of developers instead of thousands

it makes sense if you need to control the slop chaos at a medium to big org. I would go further and add controls like component libraries and packages (shadcn/react query) and have rules to always use them so the code has some natural guardrails and is more likely to b standardized.

24 hours ago, I posted this article about @FactoryAI's take on React's useEffect which has surpassed 1.7M views and is still growing. This caught the attention of numerous execs, startup founders, and even the React core team. It has sparked a conversation about a paradigm shift in how we design software for the agentic era. Traditionally, software frameworks were designed for humans who spent time mastering fundamentals before writing their first line of code. Today, that is no longer the case. At Factory, all of our "backend engineers" ship frontend code. Any engineer should be able to prompt agents to tweak features "out of the box" with built-in guardrails. We learned the hard way that when agents write nearly all the code, useEffect often becomes the culprit behind systemic frontend bugs. We only encountered these issues because we are constantly pushing the boundaries of agentic software development. Fixing the process is more important than fixing the (direct) problem. On a note regarding marketing strategy: traditional, polished product announcements from PR teams don't work anymore. Sharing raw, authentic, "on the ground" stories about the interesting problems teams are solving is far more engaging.

@tbpn @carrynointerest this guy sounds like an idiot secret sauce lmfao - 'ooh i have a special prompt to get the powerpoint color just right'

.@carrynointerest says Manus is one of the most underrated AI products today, and that it was a genius acquisition by Zuck: “There are simply things that you can do with Manus that no other inference provider or product can do. I would reveal them, but that’s alpha for me.” “What Zuck realized is — 'I don’t have to care about inference or who’s giving it to me, because [Manus] figured out some really special stuff around how an LLM processes data in and around a web browser.'” “And now he’s already integrating it with Ads Manager. It’s a great bet.”

@IndieGameJoe It would be interesting if we get a whole genre of games which are basically simulations of normal things people did 30 years ago (talk to stranger, go to library/bookstore, typist)

Two indie devs made a game where you run your own video store in the early 90s. It’s currently the #5 top-selling game on Steam. - Rent out VHS tapes & manage customers - Charge Late & Broken Fees - Upgrade & customise your store It’s called Retro Rewind - Video Store Simulator

This is what frustrates me. Image models are so jagged based on the prompts. Below is a comparison of Nano Banana 2 (Flash) v Flux Klein 9B. You would think the former would be so much better but it gave me imo a much worse result. Here was my prompt: Wide shot of a cramped space station tavern interior, low riveted metal ceiling with hanging amber lanterns. Eight spacemen in oil-stained gray and brown coveralls are on their feet around rough wooden tables, fists raised high gripping ceramic mugs, mouths wide open mid-shout, singing in unison. Dark ale sloshes from the mugs. The wooden tables vibrate from fist impacts. At the center of the group, a blind old man in a patched navy-blue spacer jacket grins wide, his battered red and silver accordion bellows fully extended across his chest, milky-white eyes bright in the amber lantern light. The energy in the room is electric, every face lit warm from below.. Style: Spielberg's Adventures of Tintin (2011) film aesthetic. Hyper-detailed stylized realism, expressive exaggerated facial features, warm golden cinematography, rich saturated color palette, painterly skin textures, filmic depth of field, Janusz Kamiński lighting.

@pitdesi @sfspacesquid tried hilo. its alright. smitten is still the best ice cream i've had so far in SF (would say ties with salt n straw).

@sfspacesquid Reporting from the mission: Hilo is way more packed than Smitten

SF dudes say taste is important and then go line up for Salt n Straw in Hayes instead of going to hometown creamery

@sfspacesquid salt n straw is way better

People shitting on Midjourney but tbh I've tried all the models and even the latest Seedream, Qwen and Nano Banana 2 still are meh in terms of pure aesthetics. I'm sure its part prompt skill issue but its 2026 it should not be that hard to get a mj v6+ cheap image without much effort.

@pzf_ai yeah this is something we've been thinking about at jellypod. our users regularly make 30-60 minute long podcasts and do well with their established audience but we want to make it easier for them to get discovered on all the big social media channels.

I see a lot of AI video creators expressing frustration with the platforms available to them right now. YouTube is where many of us post longer work, but a growing number of creators seem to be finding that their reach there isn't what it used to be. Most social channels are built for quick hits and virality, not for sitting down and watching a longer piece of work. Narrative films, music videos, surreal experiments; that kind of content doesn't seem to have a natural home yet, where creators can build and manage their profile. If a platform existed that was built specifically for discovering and watching longer form AI video, what would matter most to you? What would make you want to spend time there? A few things I keep coming back to: ▫️the ability to find work by mood or genre ▫️a way to follow creators whose artistic voice you connect with ▫️space for longer pieces that don't get punished by an algorithm designed for short clips Whether you make AI video or just enjoy watching it: what would pull you in?

@TheCrashlander @TheRundownAI Thanks! Here you go!

@deepwhitman @TheRundownAI This is fantastic! Can you do The Green Hills of Earth by Heinlein as a sea shanty? That would be awesome.

Someone used Suno AI to generate a Japanese metal band called Neon Oni. Fake member bios, AI-generated music videos, "Based in Tokyo" on Spotify. 80,000+ monthly listeners. Fans had it in their Spotify Wrapped top 5. Merch was selling. Then, community sleuths exposed it. Traced the creator's account to Europe. Spotted AI-generated hands in the music videos. The creator's response? Recruit 7 real musicians from actual Tokyo bands to perform the AI-generated songs live. They've now played several live shows and have more on the books. From an interview with the band's creator: "In an age where AI is taking everyone's jobs, this has actually created jobs. It's done the complete opposite." The AI --> real band transformation is a wild one.