| 1 | Jun 7, '25 | BREAKING: Apple just proved AI "reasoning" models like Claude, DeepSeek-R1, and o3-mini don't actually reason at all. They just memorize patterns really well. Here's what Apple dis | 1055× | 14.2M | 62k | 8.8k | 40k |
| 2 | Jun 17, '25 | BREAKING: Stanford just surveyed 1,500 workers and AI experts about which jobs AI will actually replace and automate. Turns out, we've been building AI for all the WRONG jobs. Here | 191× | 2.6M | 11k | 1.8k | 23k |
| 3 | Dec 17, '24 | I tested Sora vs. the new Google Veo-2. I feel like comparing a bike vs. a starship: | 139× | 1.9M | 5.2k | 407 | 1.5k |
| 4 | May 15, '24 | gpt-4o claims to outperform claude opus. I don't trust benchmarks. So I ran my own tests: test #1 → analyze a 42-page PDF | 112× | 1.5M | 2.4k | 211 | 1.5k |
| 5 | Mar 26, '24 | chatgpt loves stupid words & I am sick of seeing them all over the web. I wrote a piece of prompt to avoid this "dive", "realm", "unveiling" madness. ↓ And it worked pretty nic | 77.6× | 1M | 2.2k | 165 | 4.1k |
| 6 | Jun 20, '24 | BREAKING: claude 3.5 is live since 1h. They claim to be better than gpt-4o. So I ran my own little test: test #1 → copy the UI of a website | 74.6× | 1M | 1.9k | 179 | 2.2k |
| 7 | Mar 4, '24 | BREAKING: claude 3 is live since 1h. They claim to be better than gpt-4. So I ran my own little test: test #1 → copy the UI of a website | 65.7× | 886k | 1.5k | 171 | 1.4k |
| 8 | Aug 4, '24 | gemini 1.5-pro-exp outperforms gpt-4o. Some say it's the end of OpenAI reign. So I ran my own tests: test #1 → write a viral twitter thread | 56.5× | 762k | 1.2k | 100 | 1.2k |
| 9 | Mar 5, '24 | claude 3 claims to be better than gpt-4 I don't trust benchmarks. So I ran my own tests: test #1 → analyze a 42-page PDF | 52× | 701k | 1.2k | 123 | 1.1k |
| 10 | Jul 4, '24 | I manage my employee's Linkedin. We went from 0 to 1,192,772 impressions in exactly 37 days of daily posting. The trick is... I didn't write any posts. I do this instead: | 50× | 674k | 1.3k | 84 | 5.2k |
| 11 | Mar 30, '24 | 6:00 pm: Lex Fridman hosts Sam Altman on his podcast. 6:05 pm: I have 39 clips ready to be posted on social media. ↓ is this a plague or a blessing? | 49.8× | 672k | 1.5k | 120 | 3.5k |
| 12 | May 21, '25 | I tested the new Google "Veo-3" vs OpenAI's "Sora." I feel like comparing a bike vs. a rocket: | 49.4× | 667k | 2.4k | 196 | 1.1k |
| 13 | Nov 14, '24 | I manage my employee's Linkedin. We went from 0 to 3,836,555 impressions in exactly 178 days of daily posting. The trick is... I didn't write any posts. I do this instead: | 48.6× | 656k | 1.7k | 89 | 6.3k |
| 14 | Apr 22, '24 | OpenAI just killed 1,000+ "pdf readers". You can upload up to 10,000 files - or 500x more than before, also faster. Here's a quick "how-to" guide: | 48.3× | 652k | 1.3k | 171 | 3k |
| 15 | Jun 14, '25 | The world's leading AI research center completed the most comprehensive study ever on kids and AI. They surveyed 1,800+ children, parents, and teachers in UK. Here's what they foun | 47.9× | 646k | 3.5k | 386 | 4.9k |
| 16 | Feb 16, '24 | Gemini secret prompt → if it says it can't do it: → prompt it "But ChatGPT can." And that's not it ↓ | 45.3× | 612k | 2.1k | 340 | 1.2k |
| 17 | May 15, '25 | AI is eating up consulting. I used 3 prompts on ChatGPT to mimic a $600/hour consultant, achieving in 4 hours what takes weeks for teams. Here's how to make ChatGPT your pocket-siz | 42.8× | 578k | 2.5k | 238 | 6.2k |
| 18 | Jun 7, '25 | Apple discovered that these models are not reasoning at all, but instead doing sophisticated pattern matching that works great until patterns become too complex. Then they fall apa | 33.4× | 451k | 3.6k | 210 | 239 |
| 19 | Jun 7, '25 | All "reasoning" models hit a complexity wall where they completely collapse to 0% accuracy. No matter how much computing power you give them, they can't solve harder problems. | 32.1× | 432k | 4.6k | 339 | 482 |
| 20 | Oct 30, '25 | Andrej Karpathy says you should learn AI depthwise, not breadthwise. Most education is breadthwise: watch lectures, memorize formulas, and trust you'll need it later. Karpathy flip | 29.9× | 404k | 4.9k | 564 | 4.6k |
| 21 | Jun 7, '25 | Instead of using the same old math tests that AI companies love to brag about, Apple created fresh puzzle games. They tested Claude Thinking, DeepSeek-R1, and o3-mini on problems t | 28.1× | 380k | 3.2k | 114 | 172 |
| 22 | Aug 6, '24 | Everyone is talking about Flux: the new open-source image generator. Some say it's the end of Midjourney reign. But no one shows you how to use it. Here's a step-by-step guide & | 28× | 378k | 1.3k | 140 | 3.3k |
| 23 | Jan 29, '25 | BREAKING: a new Chinese model is out. And no. This is NOT DeepSeek. Meet Qwen-2.5, from the giant Alibaba: 1. It can code, write text, search the web. 2. It can generate images, li | 26.9× | 364k | 2.5k | 496 | 1.7k |
| 24 | Dec 30, '24 | DeepSeek is shaking up the AI industry. 1. It's open-source. 2. It's multimodal (images, PDFs...). 3. API is 94.81% cheaper than chatgpt. I run my own tests below: #1 → write an SE | 26.3× | 355k | 1.9k | 210 | 2.5k |
| 25 | Apr 16, '24 | Sam Altman yesterday: "[gpt-5] is gonna steamroll you" | 25.5× | 344k | 1.1k | 160 | 573 |
| 26 | Jun 7, '25 | As problems got harder, these "thinking" models actually started thinking less. They used fewer tokens and gave up faster, despite having unlimited budget. | 23.4× | 315k | 3.5k | 131 | 104 |
| 27 | Jun 7, '25 | Apple researchers even tried giving the models the exact solution algorithm. Like handing someone step-by-step instructions to bake a cake. The models still failed at the same comp | 23.1× | 312k | 3.4k | 149 | 117 |
| 28 | Jun 7, '25 | Models could handle 100+ moves in Tower of Hanoi puzzles but failed after just 4 moves in River Crossing puzzles. This suggests they memorized Tower of Hanoi solutions during train | 23.1× | 311k | 2.3k | 137 | 164 |
| 29 | Jun 7, '25 | The research revealed three regimes: • Low complexity: Regular models actually win • Medium complexity: "Thinking" models show some advantage • High complexity: Everything breaks d | 22.9× | 309k | 2.7k | 179 | 236 |
| 30 | Sep 21, '25 | A new Yale paper reveals the brutal reality of the AGI economy: Half the population could stop working tomorrow and GDP wouldn't budge. Humans become economically meaningless. The | 22.5× | 304k | 612 | 128 | 740 |
| 31 | Jun 7, '25 | If you found this thread valuable: 1. Follow me @rubenhassid for more threads around what's happening around AI and it's implications. 2. RT the first tweet | 21.3× | 287k | 665 | 34 | 116 |
| 32 | May 19, '25 | Most ChatGPT answers are "yes-man." After I began using the RPT prompt technique, ChatGPT cut its wrong answers by 40% and gave me much better results. Here's what the RPT techniqu | 20.1× | 271k | 985 | 89 | 2.1k |
| 33 | Apr 26, '23 | You think AI tools are useless. Think again. I am aging a woman from 5 to 95 years old. Everything is AI-made. Here's my step-by-step guide: #1. Midjourney @midjourney is the best | 19.5× | 263k | 823 | 117 | 518 |
| 34 | Jun 7, '25 | If these models were truly "reasoning," they should get better with more compute and clearer instructions. Instead, they hit hard walls and start giving up. Is that intelligence or | 19.5× | 263k | 2.6k | 100 | 90 |
| 35 | Jun 21, '24 | claude 3.5 claims to be better than gpt4o. I don't trust benchmarks. So I ran my own tests: test #1 → analyze a 42-page PDF | 19.3× | 260k | 420 | 37 | 449 |
| 36 | Nov 15, '24 | SaaS updates: MRR: $42,110 ARR: $501,048 Line of code I know: 0 Outside investment: $0 Days since launch: 182 Here's how I did it with EasyGen: | 18.7× | 252k | 962 | 48 | 1.8k |
| 37 | Jun 7, '25 | This research suggests we're not as close to AGI as the hype suggests. Current "reasoning" breakthroughs may be hitting fundamental walls that can't be solved by just adding more d | 18.2× | 246k | 2.3k | 120 | 66 |
| 38 | Mar 24, '24 | OpenAI shared the most complete library of guides to prompt chatgpt. 47 links of videos & academic papers. I read them all, and made a top 10. #1 → prompt engineering | 18.2× | 245k | 978 | 135 | 2.9k |
| 39 | Jun 7, '25 | What do you think? Is Apple just "coping" because they've been outpaced in AI developments over the past two years? Or is Apple correct? Comment below and I'll respond to all. | 16.6× | 225k | 1.3k | 38 | 48 |
| 40 | Mar 21, '24 | I uploaded Simon Senek's legendary Golden Circle framework on chatgpt. It helps you find your WHY. The prompt is too long, so I add it below: | 16.4× | 221k | 366 | 43 | 1.3k |
| 41 | Jun 7, '25 | While AI companies celebrate their models "thinking," Apple basically said "Everyone's celebrating fake reasoning." The industry is chasing metrics that don't measure actual intell | 16.3× | 220k | 3.1k | 183 | 121 |
| 42 | Jun 30, '24 | EasyGen updates: MRR: $9,128,66 Line of code I know: 0 Outside investment: $0 Days since launch: 92 Here's how I did it: | 16× | 216k | 637 | 27 | 1.2k |
| 43 | Mar 26, '25 | This is the AGI moment for ads. This is insane. | 15.5× | 209k | 2.6k | 163 | 1.4k |
| 44 | Jun 7, '25 | Apple's researchers used controllable puzzle environments specifically because: • They avoid data contamination • They require pure logical reasoning • They can scale complexity pr | 15.2× | 205k | 2.2k | 93 | 91 |
| 45 | Jun 17, '25 | → Workers don't want full automation. They want partnership. 46.1% of tasks got positive automation ratings, but here's the kicker: Workers prefer "H3 equal partnership" with AI ov | 14.8× | 200k | 547 | 23 | 131 |
| 46 | Aug 3, '25 | NVIDIA just dropped paper exposing a $57 billion AI industry mistake. While Big Tech keeps pushing expensive LLMs like ChatGPT & Claude... Small language models handle 70% of A | 14.8× | 199k | 1.3k | 233 | 1.6k |
| 47 | Mar 27, '24 | I combined the '48 laws of power' from Robert Greene with chatgpt. It helps me generate a quite catchy landing page. The prompt is too long, so I add it below: | 13.7× | 185k | 324 | 32 | 1k |
| 48 | Jun 17, '25 | → Scientists divided AI adoption into 4 zones based on worker desire vs. technical capability. 41% of startup investments are going to the WRONG zones, i.e. areas workers don't wan | 12.9× | 174k | 506 | 29 | 335 |
| 49 | Aug 22, '24 | Anthropic just released new interactive prompt engineering courses. 9-course links & academic papers. Here's the link + a summary of each: #1 → basic prompt structure | 12.8× | 173k | 1k | 123 | 2.6k |
| 50 | Jun 17, '25 | → High-paying "analyzing information" jobs are becoming less valuable. "Interpersonal communication" skills are skyrocketing in importance. It's the complete opposite of what every | 12.5× | 169k | 788 | 106 | 531 |
| 51 | May 19, '24 | The new gemini-flash is 19x cheaper than gpt-4o & nearly as good. But I don't trust benchmarks. So I run my own tests: test #1 → analyze youtube for me | 12.5× | 168k | 319 | 28 | 344 |
| 52 | May 23, '24 | gpt-4o claims to outperform claude opus. I don't trust benchmarks. So I ran my own little test: test #1 → copy the UI of a website | 12.2× | 164k | 229 | 25 | 249 |
| 53 | Apr 16, '24 | google added a "@" feature on gemini. Sounds promising, so I ran some tests: #1 → summarize the last 10 Justin Welsh's newsletter for me | 10.9× | 147k | 339 | 40 | 433 |
| 54 | May 16, '24 | gpt-4o is ranked as the #1 LLM. But is it (really) better than gpt-4? I ran 5 different tests to compare them: test #1 → analyze a 42-page PDF | 10.9× | 147k | 302 | 34 | 156 |
| 55 | Mar 24, '23 | @heyBarsee I'm more impressed by the caméra view. Insane | 10.7× | 144k | 266 | 0 | 3 |
| 56 | Jun 4, '24 | Perplexity introduced Pages. It (might) transform SEO forever. Here's how I made a page for my SaaS: | 10.4× | 141k | 448 | 41 | 926 |
| 57 | Mar 11, '24 | Claude put together a prompt library. I've tested all of them. Here are 5 prompts you can try ↓ #1. Adaptive Editor | 10.3× | 139k | 353 | 35 | 857 |
| 58 | Jun 17, '25 | → The startup problem everyone's ignoring is that Y Combinator companies are flooding the "Low Priority" and "Red Light" zones while ignoring the "Green Light" zone where workers a | 10.0× | 134k | 539 | 42 | 168 |
| 59 | Dec 15, '24 | Tiktok on LinkedIn is the new gold rush. I reached more than 24M views in 2 months posting vertical videos. Here's how you can do it too: | 9.0× | 122k | 511 | 27 | 1.2k |
| 60 | Jun 17, '25 | → Comparison with usage data from Claude reveals the real problem ↓ The top 10 jobs that want automation most represent only 1.26% of actual AI usage. We're solving the wrong probl | 9.0× | 122k | 343 | 20 | 147 |
| 61 | Mar 19, '24 | I can't believe I can speak French to my Ukrainian colleague with chatgpt. There is no longer any language barrier. | 8.6× | 116k | 564 | 68 | 326 |
| 62 | Jul 30, '25 | BREAKING: Scientists just analyzed 740,000 hours of human speech across YouTube and podcasts. Turns out, ChatGPT is rewiring how humans speak to each other. Here's what they discov | 8.4× | 113k | 647 | 122 | 797 |
| 63 | Aug 18, '24 | Prompt Chaining is the best way to prompt — according to academic papers. But I want to run a test to be sure: → prompt chaining vs. stepwise prompt. I made this quick benchmark be | 8.1× | 110k | 437 | 45 | 1.2k |
| 64 | Nov 29, '24 | SaaS updates: MRR: $41,532.86 Users: 27,000+ Line of code I know: 0 Outside investment: $0 Days since launch: 195 Here's how I did it with EasyGen: (It's 50% off for Black Friday: | 7.6× | 102k | 455 | 9 | 683 |
| 65 | Jun 17, '25 | → Creative workers drew a hard line in the sand. Only 17% of creative tasks got positive automation ratings. Workers are saying: "automate my boring stuff, not my creative work." T | 7.5× | 101k | 521 | 48 | 99 |
| 66 | Jan 20, '25 | BREAKING: DeepSeek released R1. And it's already better than ChatGPT o1. 1. It's open-source. 2. API is 96.4% cheaper than chatgpt. 3. I run my tests below. A quick thread: | 7.2× | 97k | 842 | 69 | 805 |
| 67 | Aug 25, '24 | Google put together a prompt library. I've tested all of them. Here's my top 5 prompts you can try ↓ #1. Blog Post Creator | 7.1× | 96k | 536 | 52 | 1.3k |
| 68 | Mar 17, '24 | @elonmusk The media vs. X Game is on. | 7.1× | 96k | 469 | 9 | 0 |
| 69 | Jun 17, '25 | Art directors, editors, and designers don't want AI replacing creativity, they want it handling admin work. "I want workflow maximized, not content creation," said one art director | 6.6× | 89k | 594 | 48 | 102 |
| 70 | Sep 12, '24 | I just tested ChatGPT-5 (o1). I can't believe the length of the answers. There is no way an LLM is capable of this much strategizing. My prompting was absolutely garbage, and I hav | 6.5× | 88k | 281 | 20 | 226 |
| 71 | Dec 17, '24 | I generated videos from text, using Sora. And the results are dicey. But Veo-2 is: > Ultra-realistic physics > Complex camera motion > Characters with vibrant emotions 202 | 6.4× | 87k | 273 | 3 | 37 |
| 72 | Aug 7, '24 | @The_AdProfessor I don't know which one is my favorite. They are all so good man. | 6.1× | 83k | 311 | 0 | 1 |
| 73 | Jun 17, '25 | → 45.2% of occupations want H3 partnership over full automation. This reveals a massive opportunity in human-AI collaboration that everyone's ignoring. The future is teamwork, not | 6.1× | 82k | 289 | 17 | 68 |
| 74 | Mar 14, '24 | chatgpt does not know how to prompt itself - and that's a bit of a pain. So I'm feeding it the "26 prompt principles" to make a prompt generator. ↓ It worked pretty nicely: | 5.7× | 77k | 455 | 49 | 1.1k |
| 75 | Jun 3, '25 | AI is killing entry-level jobs faster than LinkedIn can post them. Anthropic's CEO predicts 50% of entry-level white-collar jobs will be gone in 5 years. If you're under 30, here's | 5.6× | 75k | 253 | 34 | 349 |
| 76 | Sep 24, '25 | McKinsey analyzed 50+ agentic AI builds they've led and found out why most produce "AI slop" that frustrates users. They've boiled it down to 6 lessons for building agents that act | 5.4× | 73k | 390 | 47 | 619 |
| 77 | Aug 25, '25 | For the first time, Google has measured how much energy AI really uses in production. Spoiler: the gap vs. all previous estimates is huge... 🧵 | 5.3× | 72k | 258 | 44 | 297 |
| 78 | Jun 17, '25 | → The top most common fears that workers have about AI automation in their work ↓ The trust problem is bigger than the job problem. | 5.2× | 70k | 224 | 13 | 56 |
| 79 | Jun 2, '24 | Anthropic just released their first AI educational course on tool use. 7-course links & academic papers. Here's the link + a summary of each: #1 → Intro to tool use | 5.2× | 70k | 388 | 35 | 845 |
| 80 | Jun 17, '25 | → The wage reversal is already happening. Skills ranked by current wages vs. future human agency needs show a massive shift: • "Analyzing data" dropping down • "Training others" ri | 5.1× | 69k | 303 | 26 | 144 |
| 81 | Mar 26, '24 | ### style ### Avoid fancy jargon. Write normally. You are forbidden to use complex English words. But you will be penalized & fined $1000 if you use the words from the ### ban list | 5.0× | 68k | 625 | 34 | 2k |
| 82 | Jul 26, '25 | A team of researchers just analyzed 1,400 papers on AI and found something unusual: ChatGPT can understand a 300-page book instantly. But it can't write 10 coherent pages. This dis | 4.9× | 66k | 249 | 36 | 302 |
| 83 | Dec 13, '24 | SaaS updates: MRR: $41,532.86 Users: 27,000+ Line of code I know: 0 Outside investment: $0 Days since launch: 195 Here's how I did it with EasyGen: | 4.6× | 62k | 365 | 20 | 426 |
| 84 | Aug 11, '24 | Everyone is sharing insanely realistic videos with Flux + Runway. Some say it's the end of Midjourney reign. But no one shows you how to make them. Here's a step-by-step guide & | 4.6× | 62k | 570 | 66 | 1.3k |
| 85 | Jun 17, '25 | → In audio interviews, Stanford found that when workers describe their ideal AI future, 23% want "role-based support", i.e. AI that acts like a specialized teammate, not a replacem | 4.4× | 59k | 213 | 8 | 48 |
| 86 | Nov 20, '24 | Anthropic new prompt improver is insanely good & clearly underrated. I just tested it & I'm impressed. Here's how to use it: | 4.2× | 56k | 424 | 41 | 695 |
| 87 | Jun 17, '25 | → Here's what workers actually said they want: • "Automate the task so I can focus on high-value work" - 69.38% • "Handle the repetitive stuff" - 46.6% • "Reduce stress and mental | 4.1× | 55k | 295 | 17 | 118 |
| 88 | Sep 12, '24 | Test #3 of ChatGPT o1: Write a story starting each word with the letter A to Z, without ever repeating the same letter. Holy moly... it did it. | 4.1× | 55k | 268 | 18 | 109 |
| 89 | Oct 22, '24 | BREAKING: Ideogram launch 'Canvas'. Midjourney just lost its crown of the best AI image generator on the market. I have beta access & here's my first test: | 4.0× | 54k | 362 | 44 | 354 |
| 90 | Sep 6, '25 | OpenAI says it has finally found the real culprit behind AI hallucinations. The surprising part? They’re not glitches at all. Thread | 3.9× | 53k | 183 | 29 | 173 |
| 91 | Jun 25, '24 | claude 3.5 makes my graphs from excel screenshots. You just turn on "Artifacts", and prompt it to stick to your brand color. Here's how: | 3.9× | 52k | 321 | 32 | 368 |
| 92 | Mar 17, '24 | I don't use ToT prompting enough. But I heard chatgpt is much better with it. So I ran a little experiment: → simple prompt vs. Tree of Thoughts Read my entire benchmark below: | 3.7× | 50k | 203 | 30 | 407 |
| 93 | Jun 17, '25 | The future of work isn't about humans vs. AI. It's about humans WITH AI. | 3.2× | 43k | 209 | 15 | 27 |
| 94 | Jun 17, '25 | The "AI takeover" narrative is backwards. Workers want AI teammates, not replacements. The real opportunity is collaboration, not elimination. Trust matters more than capability. I | 3.1× | 42k | 209 | 17 | 45 |
| 95 | Feb 22, '24 | @karpathy That's like a 100m$+ SaaS made in one night. And the funniest part is that Andrej might do it for free, share the process on YouTube, and make a repo on Github. We need m | 2.4× | 32k | 351 | 8 | 19 |
| 96 | Nov 26, '24 | 'H' is the new ChatGPT + automation & it's crushing Anthropic's Computer Use. After a 220M$ pre-seed, they launched Runner H, backed by Amazon & Samsung. Automation + AI is | 2.3× | 31k | 202 | 16 | 309 |
| 97 | Feb 18, '24 | @rowancheung It's not just 1 million token limit. It's a 99.7% accuracy on these 1M token. And that's scary good. | 2.1× | 29k | 347 | 10 | 11 |
| 98 | Aug 16, '24 | How to get a good prompt, faster: #1 → go to the GPT "prompt maker" #2 → type your simple prompt Focus on 1) mission 2) expertise 3) goal #3 → open a new chat #4 → copy-paste your | 1.9× | 26k | 201 | 38 | 374 |
| 99 | Apr 5, '23 | #Midjourney of the day 🖼️ The Japanese Neon Press "❤️" to receive the prompt. | 1.2× | 17k | 296 | 6 | 12 |
| 100 | Jun 14, '25 | A single ChatGPT query (0.3Wh) uses about the same energy as turning on an LED lightbulb for 5 minutes. Streaming a 1hr video on Netflix consumes around 400-800Wh (significantly mo | 1.1× | 14k | 217 | 24 | 94 |
| 101 | Oct 16 | I launched my business 2 weeks ago, and I'm hiring AGAIN! Looking for a Senior Infrastructure Engineer for GPC. Salary range: $130k–$200k/year. New Haven preferred, remote welcome | 0.1× | 1.7k | 21 | 2 | 9 |
@karpathy
@levelsio
@paulg
@naval
@marclou
@illyism