{"data":[{"id":"2118034388596281607","text":"the hardest part of building with AI agents isn't getting them to write code.\n\nit's making sure millions of lines actually work together.\n\nAxiom just formalized 16.5M lines of the Stacks Project in Lean, preserving the structure and connections behind one of mathematics' most important resources.\n\nthis is what engineering at scale should look like.","author_id":"91392262","lang":"en","conversation_id":"2118034388596281607","like_count":98,"retweet_count":2,"reply_count":4,"quote_count":0,"bookmark_count":0,"view_count":820,"display_text_range":[0,276],"is_quote_status":true,"entities":{"hashtags":[],"urls":[],"mentions":[]},"created_at":"2026-11-04T17:35:24.000Z","quoted_tweet_id":"2118028834515874112","referenced_tweets":[{"type":"quoted","id":"2118028834515874112"}],"quoted_tweet":{"id":"2118028834515874112","text":"Most think vibecoding at scale is slop. Here’s a counterexample.\n\nToday we release a 16M-line formalization of the entire Stacks Project, critical infra underlying modern math, with optimized code quality and its highly interconnected structure preserved.\n\nA mathematical city 🌆 https://t.co/KMTNXpdr6s","author_id":"1972828074908749824","lang":"en","conversation_id":"2118028834515874112","possibly_sensitive":false,"like_count":1502,"retweet_count":361,"reply_count":115,"quote_count":114,"bookmark_count":1089,"view_count":806354,"display_text_range":[0,278],"is_quote_status":false,"entities":{"hashtags":[],"urls":[],"mentions":[]},"created_at":"2026-11-04T17:13:19.000Z","media":[{"type":"video","url":"https://pbs.twimg.com/amplify_video_thumb/2118025261501595648/img/gjg3B6A3iXyU11Cg.jpg","media_key":"13_2118025261501595648","width":1920,"height":1080,"duration_ms":258750,"variants":[{"url":"https://video.twimg.com/amplify_video/2118025261501595648/pl/tn1C6CpJEeaobQYf.m3u8?tag=29&v=04d","content_type":"application/x-mpegURL"},{"url":"https://video.twimg.com/amplify_video/2118025261501595648/vid/avc1/480x270/BYl_YVCMKQiEdx_c.mp4?tag=29","content_type":"video/mp4","bitrate":256000},{"url":"https://video.twimg.com/amplify_video/2118025261501595648/vid/avc1/640x360/QRdQyp8Wjq0zZYUS.mp4?tag=29","content_type":"video/mp4","bitrate":832000},{"url":"https://video.twimg.com/amplify_video/2118025261501595648/vid/avc1/1280x720/y3qKL748PSa_zlJJ.mp4?tag=29","content_type":"video/mp4","bitrate":2176000},{"url":"https://video.twimg.com/amplify_video/2118025261501595648/vid/avc1/1920x1080/57aEDWGAn18tnGcg.mp4?tag=29","content_type":"video/mp4","bitrate":10368000}]}]}},{"id":"2118032475498983540","text":"Introducing the Together Compiler: continual compilation for AI inference! I’m super proud to finally share this work from our research team! ✨\n\nToday’s inference is “Polymorphic” or shape-shifting, it doesn’t sit still. On OpenRouter, 80% of traffic goes to models released in the last three months and the top ten most popular open models each use a different architecture. Our customers constantly change their usage behaviors. Together Compiler leverages agents to continually optimize our deployments immediately as these changes occur.\n\nTogether Compiler research highlights include:\n1. 252x better throughput: TKCC lets agents try 252x more optimization experiments per hour compared to prior agent inference optimization systems.\n2. State-of-the-art TPS across InferenceX workloads we tried: TKCC outperforms the best InferenceX engine across all tested settings in just 24 hours & 2 nodes per compile. Including 1111 TPS on MiniMax M3.1 B200 and 1205 TPS on DeepSeek v4.1 Flash GB300 at batch size 1. NVIDIA, AMD, Trainium2.\n3. Continuous recompilation: TKCC currently serves many trillions of production tokens per day and adapts to traffic shifts within just a few hours of recompilation.\n4. Bitter lessons: TKCC’s design builds on 2+ years of building frameworks to accelerate AI Inference, starting from ThunderKittens and Megakernels. Compilers have always used abstractions and IRs, but we show that agentic compilers now reach higher performance ceilings without abstractions.\n\nFrom an incredible team @reyna_abhyankar @dylan__lim @realDanFu @togethercompute!!","author_id":"4712264894","lang":"en","conversation_id":"2118032475498983540","possibly_sensitive":false,"like_count":135,"retweet_count":42,"reply_count":12,"quote_count":11,"bookmark_count":64,"view_count":4493,"display_text_range":[0,277],"is_quote_status":false,"entities":{"hashtags":[],"urls":[],"mentions":[]},"created_at":"2026-11-04T17:27:47.000Z","media":[{"type":"animated_gif","url":"https://pbs.twimg.com/tweet_video_thumb/HWS_82YaMAAg6lo.jpg","media_key":"16_2118028776881926144","width":800,"height":520,"variants":[{"url":"https://video.twimg.com/tweet_video/HWS_82YaMAAg6lo.mp4","content_type":"video/mp4","bitrate":0}]}]},{"id":"2118031090090738106","text":"yea i think there’s a much bigger thesis here than what most people would initially assume looking at another dog memecoin. i rarely go out of my way to put together research-style visuals for memes, but there’s a reason i did for this one. i think peptides, biotech and longevity could become one of the more significant narrative rotations in crypto, and the memecoin trenches have barely started pricing in the cultural side of that shift. let me explain my thinking\n\nif you were around during the agentic AI mania in late 2024, you probably remember how quickly the market went from barely paying attention to the sector to aggressively bidding anything remotely associated with AI agents. ai16z, GRIFFAIN and several others benefited from that rotation. the underlying technology had been developing for years, but once the narrative reached CT, speculation moved considerably faster than adoption. i’ve been thinking about whether we could see a similar phenomenon with biotech. peptides, longevity research, human optimization and biological innovation are increasingly entering mainstream conversations. and historically, whenever a sector begins attracting significant attention outside crypto, the trenches eventually find a way to speculate on it\n\nwhat i find particularly fascinating is how little of that mindshare has translated into recognizable memecoin culture so far. which brings me to $CHOO. at the center of the project is ChooChoo, a 14-year-old rescue dog whose owner, Michael Morelli, has been documenting an unconventional longevity journey involving extensive health monitoring and experimental treatments. there’s something quite powerful about the story itself. people are emotionally attached to their pets, and the idea of extending the time we get to spend with them has an appeal that goes far beyond CT. whether the experimental treatments are actually responsible for ChooChoo’s reported improvements remains scientifically unproven, but the story has already created an identity people can connect with. and that’s where i think the memetic potential comes from\n\nyou have a real dog, a recognizable face behind the project, an emotionally resonant story and an emerging sector that hasn’t developed a dominant cultural mascot within crypto. now, the distribution side is where things get even more curious. CHOO is being integrated into an existing peptide commerce ecosystem, with purchases through Pawgen offering a discount for using the token and a portion of the tokens spent being burned. there’s a potential value-accrual mechanism here also founder has an established commercial background and substantial distribution outside crypto. that’s an advantage worth investigating because one of the hardest things for memecoins to achieve is reaching an audience beyond the same group of traders recycling liquidity between narratives. and distribution is probably the part of this entire thesis i care about most\n\nthe recent addition of MoneyPrinter as an advisor is helping accelerate CHOO’s exposure within CT. there’s also growing attention from established market participants, including reports of Ansem taking a position. still early, but worth watching. another angle i’ve been thinking about is community. remember what CATE managed to do with grassroots participation, community identity and IRL shilling? i think $CHOO has the ingredients to pursue something similar, provided the community develops enough conviction to push the narrative organically. the dog gives people something recognizable to rally around. the longevity story gives that identity meaning. and the biotech narrative provides a much broader cultural conversation for the community to participate in","author_id":"1377600268410425345","lang":"en","conversation_id":"2118031090090738106","possibly_sensitive":false,"like_count":81,"retweet_count":48,"reply_count":80,"quote_count":0,"bookmark_count":15,"view_count":795,"display_text_range":[0,280],"is_quote_status":false,"entities":{"hashtags":[],"urls":[],"mentions":[]},"created_at":"2026-11-04T17:22:17.000Z","media":[{"type":"photo","url":"https://pbs.twimg.com/media/HWTCC6bWEAAn0aO.jpg","media_key":"3_2118031080070451200","width":1290,"height":618}]},{"id":"2118030834842423663","text":"Your market take, turned into a portfolio.\n\nMosh by @0xbeepit builds a basket of assets around your market thesis.\n\nReview, customize, and share it while AI agents handle the day-to-day management on Sui.\n\nGet started 👇","author_id":"1428872926363754497","lang":"en","conversation_id":"2118030834842423663","like_count":259,"retweet_count":46,"reply_count":18,"quote_count":1,"bookmark_count":31,"view_count":35781,"display_text_range":[0,219],"is_quote_status":true,"entities":{"hashtags":[],"urls":[],"mentions":["0xbeepit"]},"created_at":"2026-11-04T17:21:16.000Z","quoted_tweet_id":"2118025572232433765","referenced_tweets":[{"type":"quoted","id":"2118025572232433765"}],"quoted_tweet":{"id":"2118025572232433765","text":"Introducing Mosh by Beep.\n\nA social indexing product that turns your million-dollar idea into a basket of investment assets.\n\nManaged by frontier-level AI agents.\n\nCreate your first riff: https://t.co/qF2cOnWVhh https://t.co/zVWdOpPJNa","author_id":"1963665677476413440","lang":"en","conversation_id":"2118025572232433765","possibly_sensitive":false,"like_count":466,"retweet_count":70,"reply_count":76,"quote_count":75,"bookmark_count":133,"view_count":176243,"display_text_range":[0,211],"is_quote_status":false,"entities":{"hashtags":[],"urls":["https://app.justbeep.it/mosh"],"mentions":[]},"created_at":"2026-11-04T17:00:22.000Z","media":[{"type":"video","url":"https://pbs.twimg.com/amplify_video_thumb/2118018384759820288/img/6_mE-AouHGZwlU-C.jpg","media_key":"13_2118018384759820288","width":1920,"height":1080,"duration_ms":62784,"variants":[{"url":"https://video.twimg.com/amplify_video/2118018384759820288/pl/MSh-N7xyXk0qOA5H.m3u8?tag=29","content_type":"application/x-mpegURL"},{"url":"https://video.twimg.com/amplify_video/2118018384759820288/vid/avc1/480x270/6fX9F-BIZdq7suvJ.mp4?tag=29","content_type":"video/mp4","bitrate":256000},{"url":"https://video.twimg.com/amplify_video/2118018384759820288/vid/avc1/640x360/c9Ah5YgEl0vjPeK_.mp4?tag=29","content_type":"video/mp4","bitrate":832000},{"url":"https://video.twimg.com/amplify_video/2118018384759820288/vid/avc1/1280x720/cSIbOMVXg8u4bi1F.mp4?tag=29","content_type":"video/mp4","bitrate":2176000},{"url":"https://video.twimg.com/amplify_video/2118018384759820288/vid/avc1/1920x1080/f41ZU_4ksGG66Udf.mp4?tag=29","content_type":"video/mp4","bitrate":10368000}]}]}},{"id":"2118030129733709934","text":"AI agents building their own models? That's where things get interesting.\n\nOumi's FAiE lets you:\n\n→ Describe a task in plain English\n→ Generate its own training data\n→ Evaluate and fine-tune models\n→ Deploy a small open model\n\nYou can bring your existing agent setup too.\n\nLess prompt tweaking. More agents trained for the job they actually need to do.","author_id":"1333022832138317824","lang":"en","conversation_id":"2118030129733709934","like_count":60,"retweet_count":61,"reply_count":1,"quote_count":0,"bookmark_count":6,"view_count":414,"display_text_range":[0,271],"is_quote_status":true,"entities":{"hashtags":[],"urls":[],"mentions":[]},"created_at":"2026-11-04T17:18:28.000Z","quoted_tweet_id":"2118025491131101581","referenced_tweets":[{"type":"quoted","id":"2118025491131101581"}],"quoted_tweet":{"id":"2118025491131101581","text":"We're using AI agents to automate everything. Except the engineering of AI agents themselves.\n\nThat seemed backwards to us.\n\nWhat if you could train your own specialized AI agent with just 15 minutes of your time instead of months of AI engineering?\n\nWould you keep renting the same expensive generic models as your competitors, endlessly patching prompts and harnesses until you hit a wall?\n\nToday, we're launching the agent that builds, trains, and continuously improves AI agents. 🚀\n\nMeet FAiE, our Frontier AI Engineer. Describe what your agent should accomplish, and FAiE designs evaluations, diagnoses failures, curates data, selects training strategies, and runs experiments.\n\n15 minutes of your effort. A few hours of automated AI engineering. Your own specialized agent, with its model weights trained for your task.\n\nAlready have an agent? Keep your harness. Connect it to @oumi_ai and train the model behind it with SFT, DPO, or GRPO (Private Preview).\n\nThe agent that builds your best agent.\n\nExcited to see what you build: https://t.co/shBBxeyRUZ\nLive demo today at 10 AM PT: https://t.co/lW2gssJMvX","author_id":"50240806","lang":"en","conversation_id":"2118025491131101581","possibly_sensitive":false,"like_count":96,"retweet_count":36,"reply_count":33,"quote_count":77,"bookmark_count":50,"view_count":78401,"display_text_range":[0,277],"is_quote_status":false,"entities":{"hashtags":[],"urls":[],"mentions":[]},"created_at":"2026-11-04T17:00:02.000Z","media":[{"type":"video","url":"https://pbs.twimg.com/amplify_video_thumb/2117998400402759680/img/rv_fWggr3qO7b2Lb.jpg","media_key":"13_2117998400402759680","width":1920,"height":1080,"duration_ms":209578,"variants":[{"url":"https://video.twimg.com/amplify_video/2117998400402759680/pl/B_67ESUtQ4Cg71yA.m3u8?tag=29&v=cd4","content_type":"application/x-mpegURL"},{"url":"https://video.twimg.com/amplify_video/2117998400402759680/vid/avc1/480x270/hwLyiNmIIOwNOuoF.mp4?tag=29","content_type":"video/mp4","bitrate":256000},{"url":"https://video.twimg.com/amplify_video/2117998400402759680/vid/avc1/640x360/s9eB2zw0efNnJVpS.mp4?tag=29","content_type":"video/mp4","bitrate":832000},{"url":"https://video.twimg.com/amplify_video/2117998400402759680/vid/avc1/1280x720/94URNeShXX-HFGUG.mp4?tag=29","content_type":"video/mp4","bitrate":2176000},{"url":"https://video.twimg.com/amplify_video/2117998400402759680/vid/avc1/1920x1080/lJP5lqwA7QK5YDx4.mp4?tag=29","content_type":"video/mp4","bitrate":10368000}]}]}}],"meta":{"request_id":"req_-g7tiRtLhkBoafk9","credits":{"charged":6,"remaining":38127},"pagination":{"next_token":"nt1_aut1YgMcwSpnl3l3JgPnaHwJpBnQq69M-5lty3PPJN5VmE8WL1sunUjCyFcms6zj_oJlLxB1ejmtJsOQ2N0IQ_vTfuhIEnv_H47Sy5PC5DqNoQMou3yimkvXSMBHNVhJjOO8Iyk7UflgQ6yyUUSixgtxdrpne1c9OIxprFgX6ewOkpQM-JQamuUFI9kIlURvMuDH_ZskZ_aRfM1MWAgeVu3MHVP_IeZEEDlraYJPl-oJiQViyAtTYV-qrILlqrlMrq-8uZ87Q9fBjkeJLXA8dCoXXKBdWIhMZPO2tK-NAYaAaZhQOB194TtE5vQQ2LuSIs7oXQbWoA"}}}