AfterDark (@After_Dark_AI) header
AfterDark (@After_Dark_AI) profile picture

AI Chaos Control Center I break new AI tools so you know which ones are worth your time And sometimes build weird things with them

Joined October 2026
Fable left the $20 Claude plan last night. So I gave Opus 5.5 one prompt and got this A whole volcanic island in one HTML file. No 3D models, no images, just code. Hit ERUPT and it blows, lava runs down the slopes and lightning flashes in the ash
6
11
223
AfterDark retweeted
There is no 3D software in this clip. The diamond ring is a few thousand points, and every one of them comes out of numpy. The band is a circle bent through space. The stone is a set of flat faces worked out with sines and cosines. Then everything rotates a little each frame and gets drawn again. A graphics engine hides all of this from you. Written out by hand, a shiny object turns out to be mostly arithmetic. Could you tell this was code if nobody showed you the bottom half of the screen?
4
2
9
89
AfterDark retweeted
Build a single-file HTML 3D scene with three.js (load it from a CDN, no other assets): "NEON COAST — Helicopter Pursuit", a cinematic police chase in an original 1980s Miami-style neon city at night. SCENE - A coastal avenue with pink/teal/purple art-deco hotels, glowing neon signs and window lights, palm trees, street lamps with orange light pools, cross streets, crosswalks, lane markings, a beach and the sea with a night sky, stars and a moon. - Moving traffic (sedans, hatchbacks, vans) and parked cars. - Everything is generated in code: use InstancedMesh for buildings and traffic, canvas-generated textures, additive glow sprites, fog, blob shadows. ACTION - A red sports coupe flees at up to 180 km/h along a fixed route with 4 sharp corners, weaving between cars and drifting in the turns. - 4 police cars with flashing red/blue sirens chase it. At a roadblock one patrol car crashes (sparks, smoke). - Near the end the coupe takes a ramp and jumps over an open drawbridge gap while the remaining cops brake and stop. CAMERA AND HUD - The camera is a helicopter filming the chase, with hard cuts between shots: a high establishing shot, behind-high, top-down at corners, a low front shot, side shots, a static shot for the jump, and a pull-up at the end. Add slight handheld shake. - Heli-cam HUD: REC/LIVE timecode, altitude and bearing, big speed readout, number of pursuit units, a circular minimap with suspect and police dots, target brackets on the car, a bottom news ticker with short English lines, and a title card "NEON COAST / HELICOPTER PURSUIT". Add a pink/orange colour grade and a vignette. - Show a small label "3D SIMULATION". Do not copy any game logos, fonts or characters, and use a generic red coupe without brand badges. TECH - Use a fixed timeline (~41 seconds) that loops, driven by a deterministic step(dt) function. Keep it smooth at 60 fps on a normal laptop. - Optional: generate synthwave music, siren and engine sounds with the Web Audio API, started by a click. Return the complete HTML file.
1
1
6
250
The hardest thing in AI video is keeping the same face from shot to shot So I gave PixVerse V6 one prompt with 4 shots and one old lighthouse keeper. First try, no edits, the sound is generated too Same woman, same braid, same coat in every shot. It did skip the sea monster I asked for in the last one
7
16
273
AfterDark retweeted
This AI video is absolutely a masterpiece 🔥 Seedance 2.5
23
14
158
11,560
AfterDark retweeted
I let six AI agents run a simulated trading desk overnight Start: $150 By 7am: $12,860 Each agent has one job: > scanner finds new pairs and volume spikes > risk check filters thin-liquidity tokens > sizer sets position from equity, never a fixed ticket > executor opens and closes, tracks slippage > logger writes the reason behind every trade > equity updates on every close The shift that got me: I stopped asking "would you long this?" and started reading the log of what already happened. Every line has a reason. That's what makes it watchable. Would you be interested in an article on how I built this terminal?
9
2
19
395
Alibaba just stopped giving away its best model In August they opened the weights of Qwen3.8-Max, 2.4T parameters and the biggest open model anyone had published at the time. Qwen4-Max stays closed. Only the smaller Qwen4 models get open weights It's cheap at $1.50 in and $4.50 out per 1M tokens. You just can't download this one Was open frontier AI just a phase?
📢 Meet Qwen4-Max, your new AI teammate for work, creativity, and discovery! Qwen4-Max is built on a new MoE architecture with native vision and a 1M-token context window. Highlights: - Outperforms Qwen3.8-Max with up to ~50% fewer tokens, while average cost per task drops over 60% in our internal evaluation. - Advances real-world productivity through greater proficiency with professional tools, stronger long-horizon execution, and higher-quality deliverables. - Handles real tasks, including 3D modeling, game development, manufacturing, finance, law, healthcare, and frontier scientific discovery. - $1.5/M tokens for input, $4.5/M tokens for output, and $0.075/M tokens for implicit caching — all lower than Qwen3.8-Max, with implicit caching down by 70%. The API is live on QwenCloud. You can also try Qwen4-Max on Qwen Studio, QwenWork and Qoder. Start building today! 🚀 - Blog: qwen.ai/blog?id=qwen4-max - API: qwencloud.com/models/qwen4-m… - Token Plan: qwencloud.com/pricing/token-… More to come! The open weights of smaller Qwen4 models will be released. Stay tuned. 👀
6
9
168
AfterDark retweeted
a full 3D fitting room in one prompt. you can build a full 3D landing page the same way. no Blender. no Spline. one prompt. i just recorded the whole process, step by step 👇
12
31
2
337
15,467
Fable 5.2 leaves the $20 Claude plan tomorrow night So I gave Fable and Opus 5.5 the same 2 big builds, first try, and checked my usage after every run Fable was 2x faster. Opus used about half the limit. Closer than I expected
13
2
14
564
An AI episode just cost $102,725 and took 45 days to make The video runs 4:37. About $370 for every second on screen @PJaccetturo posted it yesterday, 3M+ views so far. The same day a fully AI trailer called NUREMBERG passed 13M AI film isn't cheap. It's a different kind of expensive
5
1
16
431
I can't f*cking believe this is just code There isn't a single 3D model or texture in it. Claude Fable 5.2 wrote every gear and every glass tentacle inside one HTML file, and it runs live in the browser Everyone's building mechanical animals that take themselves apart. So I gave mine a button nobody else has: JELLY One click and the glass and chrome turn into gummy candy, while the gears keep spinning inside. Then it blows apart into hundreds of jelly pieces and snaps back together
14
1
27
908
GTA 6 is out next week. I asked Gemini for a GTA-style game in one prompt to see how close we are. It built a whole city with traffic, a GPS minimap, a speedometer and delivery missions with a timer. All in one HTML file, first try. Then I failed my very first delivery. $0 earned. Still a long way from GTA. But a year ago this prompt gave you a square moving around a box.
5
12
157
ChatGPT now tells businesses how many times it mentioned them. New today, in beta: Business Hub. Any eligible business in a country where ChatGPT works can claim its profile and see: → how often ChatGPT mentions you → how many people clicked through to your site → "discovery topics", the themes of conversations where your site gets cited You can also update your details and add products with prices and stock, so ChatGPT works from your info instead of guessing. If you sell anything online, claim it this week. 1.2 billion people use ChatGPT every week, and right now it describes your business without you.
5
11
129
One prompt on Fable 5.2 burned about 6x more of my Claude limit than the same prompt on Opus 5.5. Same task, Medium effort, Pro plan: Opus moved my session meter 1%, Fable moved it 6%. Both built a working Snake game. If you're on Pro: Fable is only included until Nov 16, capped at half your weekly limit. After that it runs on paid usage credits. Anthropic's own advice is to start with Opus 5.5 and save Fable for the hardest tasks. Fable for the problem you're stuck on. Opus for everything else.
8
15
181
Zoom into "I'm not a robot". It never ends. That checkbox exists to catch machines. Machines now pass it better than we do: in a 2023 UC Irvine study with 1,400 people, humans solved distorted-text CAPTCHAs 50 to 84% of the time. Bots got almost 100%, in under a second. So the video follows a machine trying to look human: captcha, server rack, a chip that looks like a city, a laptop at night, 99% buffering, 1% battery, Snake, a 3D hand with one finger too many. Then back to the fresco. From the first frame to the loop you zoom in about 500 billion times. The fresco is the real Michelangelo, every other scene is drawn in code.
5
2
15
529
ChatGPT can make GIFs now. So I tested if it understands physics. 4 classic animation tests, first try each, every one in a fresh chat: → bouncing ball: it slows down as it falls. Gravity should speed it up → Newton's cradle: I asked for 5 balls, it drew 6 → clock: 10 ticks of one second each. It froze, then jumped 3.5 seconds → pouring water: nailed it 1 out of 4. It nails how things look. How they move is still hit or miss.
8
1
18
478
Golf uses 2.5x more water than every data center in US. Power plants included. → golf courses: ~600B gallons a year (last national survey, 2013) → data centers, on site: 17 to 19B gallons (2024) → plus the water used to generate their electricity: ~230B Most "AI is draining our water" posts skip the power plants. Most "it's basically nothing" replies skip them too. Add them in and golf still wins easily. The gap is closing though: on-site use alone is projected to hit 60 to 110B gallons by 2030. Tanks are drawn to scale.
3
10
237
Before AI, this was the whole pipeline. One cutout at a time
Introducing Higgsfield Miracle. Our latest proprietary post-production technology vectorizes any video into layers for localized, pixel-level editing in real time, with world knowledge built in. Every frame holds the image depth needed for cinema-grade post-production. Now in private beta with select testers. Coming soon to Higgsfield MCP. Set up Higgsfield MCP in Claude to be the first to try it.
5
1
15
315
AfterDark retweeted
what if we could actually watch how the brain thinks memory, language, movement, attention. none of them live in one spot, they all run on connected networks and researchers are already using models to study brain scans at scale. one 2026 study (BrainIAC, nature neuroscience) was tested across 48,965 brain MRI scans it wont replace neurologists. but it might spot patterns we keep missing which part of the brain would u explore first?
11
1
18
467