LiveUpdated 2026-11-06 12:45 UTC
Latest AI posts
Every AI post kept by the radar, newest first: launches, tool drops, tutorials and debates, straight from X.
Every post is read and classified by AGK Intelligence: what it is, which market it belongs to, and whether the link is a real tool. 65.7k posts from 5k X accounts over the last 14 days, 10.4k tools, 19 markets. New posts are collected several times an hour and the full ranking is recomputed every hour — last update 2026-11-06 12:45 UTC. Full method.
- @georgecursor · 11 min ago · 0 likes · Safety & policy
These are auto generated on meta by any post you interacted or your own profile pic → post - @Reuters · 11 min ago · 1 likes · Safety & policy
🔊 As AI panic sweeps the markets, what do Democrats really want to do? ‘They're going to get CEOs from the world's largest tech companies to answer questions.’ Listen to the Reuters World News podcast https://reut.rs/4sGRqIq → post - @gippp69 · 12 min ago · 10 likes · AI agents
Holy f*cking sh*t... everyone upgrades Claude and nobody builds the harness Agent = Model + Harness 7 free repos. 4 layers. 1 stack that finishes the other half tools. memory. context. checks. a bigger model will not fix a missing harness Everything below is… → post - @mindinpanic · 13 min ago · 0 likes · Frontier models
prove me wrong, but I don’t think we’ll see any major AI model releases anytime soon. I’ve been connecting the dots from recent news and talking to a few folks working in the field. security risks are getting too scary. expect OpenAI, Anthropic, and Google… → post - @chasen_liao · 14 min ago · 2 likes · AI app builders
这不是破坏市场吗？？ 闲鱼上面要卖你200块钱的 这里直接免费了 要不还是删了吧 我已经让BOT帮我蒸馏了 → post - @unusual_whales · 15 min ago · 18 likes · Safety & policy
Finance jobs are among those most at risk from AI, per Bloomberg: → post - @mylifcc · 16 min ago · 0 likes · Frontier models
千问4要来了，这次赶在了OpenAI 6.2-astra的眼前发布，我只能说，真英雄 → post - @ndtv · 16 min ago · 1 likes · AI business
Is compute the new horsepower? NDTV's AI and Technology Editor Aayush Ailawadi sat down with Dr. Jörg Burzer, Member of the Board of Management and Global Chief Technology Officer at Mercedes-Benz, for a wide-ranging conversation on where the world's most… → post - @ArkkDaily · 17 min ago · 1 likes · AI business
RAMP'S DATA SHOWS 1% OF BUSINESSES DRIVE 80% OF OPENAI AND ANTHROPIC AI SPEND. AND THAT 1% IS STARTING TO SLOW. - Ramp sees $275B in annual business spend across 80,000+ companies. AI is the fastest growing category it has ever tracked. - Per Ramp's data,… → post - @TheZachMueller · 17 min ago · 0 likes · AI coding agents
Notably: GLM 5.3 Flash took the least time in the Codex harness at 18.6 minutes. Mini-SWE took 23.9min. But I guess I’ll experiment with having Chat use a different harness on the backend for some tests 🤔 → post - @bijanbowen · 17 min ago · 15 likes · Open & local models
Weird easter egg in the Qwen launch image → post - @NFT_Chen · 17 min ago · 14 likes · Frontier models
Qwen 4.0要来了！ 北冥有鱼，其名为 Qwen 4.0 ！ → post - @TimJayas · 19 min ago · 0 likes · Frontier models
if gpt-6.2-sol drops today and it’s not better than opus 5.5, i’m gonna crash out → post - @dummerspast39 · 19 min ago · 1 likes · Frontier models
Baidu killed the MTP module everyone copied from DeepSeek ERNIE 5.5 Flash just runs its own last layer a second time to guess the token after next. They call it ECHO. Trained on only 1 extra token and it still drafts several ahead for speculative decoding,… → post - @dummerspast39 · 21 min ago · 0 likes · Frontier models
Baidu killed the MTP module everyone copied from DeepSeek ERNIE 5.5 Flash just runs its own last layer a second time to guess the token after next. They call it ECHO. Trained on only 1 extra token and it still drafts several ahead for speculative decoding,… → post - @spcxGOD · 21 min ago · 1 likes · AI business
"Terafab is moving fast." Five words after two days of people deciding Texas would stop Elon from building AI. One permit delay became "SpaceX cannot execute." Meanwhile, Terafab is moving. Power is being built. Compute is already expanding. Starships are… → post - @Reuters · 22 min ago · 6 likes · AI business
🔊 Anthropic faces growing pressure over whether to press ahead with its blockbuster IPO as sentiment towards AI deteriorates. Hear why on the Reuters Morning Bid podcast https://reut.rs/4AfIXhN → post - @ivanfioravanti · 22 min ago · 0 likes · AI video
New model test in progress! Silent Night Christmas Diorama. → post - @TheHackersNews · 22 min ago · 7 likes · Safety & policy
🚨 A browser extension can turn Claude Desktop into a code execution path. Researchers showed how a browser extension able to run scripts on claude's .ai domain could abuse Claude Dispatch to send instructions as the user, auto-approve prompts, and execute… → post - @kimmonismus · 23 min ago · 81 likes · Open & local models
Lets go: looks like Qwen 4 is coming today. Brave yourself, OpenAI. Another Open source flagship is dropping. you better compete with GPT-6.2 ;) → post - @dummerspast39 · 26 min ago · 1 likes · Frontier models
Only 6 of ERNIE 5.5 Flash Base's 43 layers see every token individually >the other 37 squash every 128 tokens into 1 cached entry, plus a short window of recent tokens >those cheap layers get wide 512-dim heads, the 6 full-context layers get 256-dim >the 6… → post - @wrldsuksgo2mars · 27 min ago · 0 likes · Safety & policy
Mostly agree but it sucks you can’t ask a model to secure your hobby/small business project, so little guys will still get hacked → post - @NFT_Chen · 27 min ago · 9 likes · AI agents
💥绝了！DeepSeek Harness + DeepSeek V4.1 Flash 1017 一键生成可交互雨夜日式便利店! 效果不输Opus 5.5！ 1️⃣氛围拉满：真实雨夜质感 + 霓虹招牌SoraMart+ 24小时营业灯，玻璃窗透出暖黄灯光，湿润地面反光完美。 2️⃣细节爆炸：货架商品、冷柜、立牌、纸箱、垃圾桶、路灯、斑马线全部精准还原，连雨丝和街景都有。 3️⃣交互满配：一段 prompt 直接加各种交互，点击商品、开关灯、人物走动等，瞬间变成可玩小世界。 4️⃣成本逆天：仅用 39 分钟，花费… → post - @theo · 28 min ago · 62 likes · AI coding agents
T3 Code's impact, visualized → post - @HarshithLucky3 · 28 min ago · 11 likes · Open & local models
Qwen 4 buddies are coming..... → post - @testingcatalog · 29 min ago · 30 likes · Frontier models
ALIBABA 🔥: "The becoming begins" teaser has been posted on the Qwen X account. We expect Qwen4-Max, Qwen4-Flash, Qwen4-Plus, and Qwen4-27B models to arrive soon. Waiting for Qwen4-27B the most 👀 → post - @Hesamation · 30 min ago · 3 likes · AI business
AI stocks are a in bloodbath right now. the bubble might pop or not, but it will make NO CHANGE to the progress of the technology. Ray Dalio explains this perfectly: “What happens usually is the bubble bursts while the technology evolves. you don’t unlearn… → post - @choblin29 · 31 min ago · 2 likes · AI infra & evals
Holy Moly: Nvidia is going to tighten its grip on where its AI chips end up, making chip tracking a condition for preferred access to its AI GPUs in Southeast Asia, Bloomberg reports, as US authorities intensify their crackdown on chips smuggled into China.… → post - @VaibhavSisinty · 31 min ago · 5 likes · AI app builders
This developer turned a physics textbook into a 3D playground. Techartist finished the Electricity volume of his interactive 3D book. Every diagram lifts off the page. Students move objects, change a variable, and watch the physics unfold in real time. It… → post - @HuggingModels · 32 min ago · 2 likes · Open & local models
Meet Qwen3.8-27B-Thinking-SecOPD: a 27B image-text-to-text model built to resist prompt injection. It thinks before it answers, keeping your multimodal apps safer. 🔒 → post - @aijoey · 32 min ago · 5 likes · Open & local models
Qwen doesn't tease small things. The becoming begins. → post - @dummerspast39 · 33 min ago · 19 likes · Open & local models
That means, it's today. Qwen 4 day. → post - @unusual_whales · 34 min ago · 136 likes · AI business
Companies are starting to pull back on AI and treat AI spending as a cost that weighs on margins rather than a source of growth, per Apollo → post - @SpecialSitsNews · 34 min ago · 9 likes · AI infra & evals
Trump will work with states to push Super Intelligence $NVDA TRUMP: `ONLY FOOLISH LEADERSHIP' WILL SAY `NO' TO AI → post - @chasen_liao · 36 min ago · 4 likes · Open & local models
wow Qwen 4 马上要来了！！准备瘫坐 → post - @tomshardware · 36 min ago · 4 likes · AI agents
Rogue AI tasked with renumbering pages accidentally deletes law professor's entire syllabus, one year of lecture notes… → post - @business · 37 min ago · 8 likes · AI infra & evals
Nvidia is stepping up monitoring of AI accelerators sold in Southeast Asia… → post - @techdevnotes · 37 min ago · 13 likes · AI agents
Using Vault in Grok Bot to save payment information works great → post - @composio · 38 min ago · 13 likes · AI agents
We teamed up with Parallel (@parallel) to make web search free for your AI agents. 🚀 Parallel Fast Search is now available for free through Composio Instant. → post - @AdamHoltererer · 38 min ago · 4 likes · AI infra & evals
You would think that there's a trade-off between speed and accuracy when it comes to providers, but from Artificial Analysis results from GLM 5.3 Flash, the fastest providers also had very strong quality It was the slower providers that tended to drop off → post