•twitter18715
listening to
timeline
18715
mentions
tracked
tracked
net sentiment
+24%
5783 positive
11540 neutral
1392 negative
volume & sentiment over time
Sep 30 – Oct 17, 2026 · Daily
sources
most discussed
—
active days
18
mentions
Showing 11801–11825 of 18715 mentions
-
• twitter 1w neutralthis is extremely powerful this is extremely powerfulview source ↗
-
• twitter 1w neutralhttps://t.co/U7BCk9zc7i https://t.co/U7BCk9zc7iview source ↗
-
• twitter 1w positiveLaMelo shows off the handles and sinks the jumper 👏 Watch @MELOD1P in his @Timberwolves preseason debut on NBA League Pass! https://t.co/cgQEmzAamq LaMelo shows off the handles and sinks the jumper 👏 Watch @MELOD1P in his @Timberwolves preseason debut on NBA League Pass! https://t.co/cgQEmzAamqview source ↗
-
• twitter 1w neutralfor my main app here’s my stats: ( I launched this 2 months ago ) -14,000 downloads -3M views -80% prof margin - 5-6 main clippers anyone can do this btw. for my main app here’s my stats: ( I launched this 2 months ago ) -14,000 downloads -3M views -80% prof margin - 5-6 main clippers anyone can do this btw.view source ↗
-
• twitter 1w neutral@davis7 Your agent is given CSS style tokens that work with your theme as well :) https://t.co/gNbDY57lkC @davis7 Your agent is given CSS style tokens that work with your theme as well :) https://t.co/gNbDY57lkCview source ↗
-
• twitter 1w negativeif anybody has tried to launch an ai character and has failed this is probably what you're doing wrong: - make sure your account is warmed up, so many people just go raw dog a new account and get pissed when they get 0 views. - engage with all of these other ai avatar accounts so your content fits right in with them -… if anybody has tried to launch an ai character and has failed this is probably what you're doing wrong: - make sure your account is warmed up, so many people just go raw dog a new account and get pissed when they get 0 views. - engage with all of these other ai avatar accounts so your content fits right in with them - dont just swap in one viral video and expect it go viral, you have to think through why this post went viral. me as a d1 scroller, for every viral video i find, i check WHY it went viral. i look through the comments what certain thing was it that triggered people. you have to go that deep - expect the worst, you see everybody else going viral in 2 days and expect to go viral as well. it does NOT work that way. i can tell you this, most of the accounts doing good right now either tried this on a different account OR have multiple accounts running. thats all for now.view source ↗
-
• twitter 1w neutralhey @stripe feature request: a "refund minus processing fee" option in the cancel subscription modal would save my day! https://t.co/425Te44o89 hey @stripe feature request: a "refund minus processing fee" option in the cancel subscription modal would save my day! https://t.co/425Te44o89view source ↗
-
• twitter 1w neutralNew: @𝚓𝚜𝚘𝚗-𝚛𝚎𝚗𝚍𝚎𝚛/𝚝𝚊𝚗𝚜𝚝𝚊𝚌𝚔-𝚜𝚝𝚊𝚛𝚝 TanStack Start apps from JSON specs Define your components and data loaders AI composes pages and layouts, with routing and SSR For website builders, CMS-driven sites and customer portals with customizable pages https://t.co/9DL8Tdtz6h New: @𝚓𝚜𝚘𝚗-𝚛𝚎𝚗𝚍𝚎𝚛/𝚝𝚊𝚗𝚜𝚝𝚊𝚌𝚔-𝚜𝚝𝚊𝚛𝚝 TanStack Start apps from JSON specs Define your components and data loaders AI composes pages and layouts, with routing and SSR For website builders, CMS-driven sites and customer portals with customizable pages https://t.co/9DL8Tdtz6hview source ↗
-
• twitter 1w positiveWe just shipped in-app visualization capabilities in T3 Code. This enables agents to build cool dynamic experiences within the thread. Huge shoutout to @davis7 for pushing us on this (and building most of it). Didn't expect to dig it so much. https://t.co/Xj8u1c2mT6 We just shipped in-app visualization capabilities in T3 Code. This enables agents to build cool dynamic experiences within the thread. Huge shoutout to @davis7 for pushing us on this (and building most of it). Didn't expect to dig it so much. https://t.co/Xj8u1c2mT6view source ↗
-
• twitter 1w neutralEvergreen Evergreenview source ↗
-
• twitter 1w positiveThe same 111GB Qwen3.8-Flash-Next model file. The same 2016 Xeon. The same single V100. But roughly twice the decode speed. Qwen3.8-Flash-Next UD-Q4_K_XL is getting around: ~20 tok/s with Strata versus ~9 tok/s with llama.cpp Same box. Same model file. Same GPU. That’s the part I find interesting. The V100 is n… The same 111GB Qwen3.8-Flash-Next model file. The same 2016 Xeon. The same single V100. But roughly twice the decode speed. Qwen3.8-Flash-Next UD-Q4_K_XL is getting around: ~20 tok/s with Strata versus ~9 tok/s with llama.cpp Same box. Same model file. Same GPU. That’s the part I find interesting. The V100 is not exactly new hardware. It’s a 2018-era Volta accelerator with 16GB of HBM2, and the host machine is running a 2016 Xeon. Yet we’re still getting usable throughput from a 111GB MoE model. Obviously, the entire model isn’t sitting inside the V100. That’s where the setup gets interesting. Qwen3.8-Flash-Next is a huge MoE model, so the inference system can keep the GPU busy with the active computation while the rest of the experts live in system RAM. You’re effectively turning the machine into a memory hierarchy: GPU HBM for the hot path. System RAM for the much larger expert pool. CPU and memory bandwidth become part of the inference engine rather than just “extra storage.” And that’s where Strata appears to be making a difference. The model file hasn’t changed. The quantization hasn’t changed. The GPU hasn’t changed. The runtime is simply doing a better job of handling this particular architecture and memory layout. ~20 tok/s from a 111GB model on a single V100 is already interesting. Getting roughly 2× the throughput of the same workload through llama.cpp is even more interesting. It reinforces something I’ve been seeing repeatedly with Qwen3.8-Flash-Next: The model’s theoretical parameter count tells you surprisingly little about how it will actually run on a given machine. What matters is: How many parameters are active? Where are the experts stored? How quickly can they move between RAM and VRAM? How efficiently does the runtime schedule them? How much computation can be kept on the GPU? How well does the inference engine exploit speculative decoding and caching? And how much overhead does the memory hierarchy introduce?view source ↗
-
• twitter 1w neutralRT @megadevhq: "JS went from a language for editing things in a page to a thing you can use to build whatever you want to make because node… RT @megadevhq: "JS went from a language for editing things in a page to a thing you can use to build whatever you want to make because node…view source ↗
-
• twitter 1w neutralhiring someone on linkedin has become equivalent to finding a bf on tinder. hiring someone on linkedin has become equivalent to finding a bf on tinder.view source ↗
-
• twitter 1w neutrali should have been called janelle monetization the way i run a nonbinary sack up i should have been called janelle monetization the way i run a nonbinary sack upview source ↗
-
• twitter 1w neutralFive of my last 30 Claude Code sessions had worked in the same worktree. One hit a git lock because another session was fetching at the same moment. Now the first session to write in a worktree claims it, and a second one has to ask me first, naming the session that got there first. https://t.co/id1VSv8mD9 Five of my last 30 Claude Code sessions had worked in the same worktree. One hit a git lock because another session was fetching at the same moment. Now the first session to write in a worktree claims it, and a second one has to ask me first, naming the session that got there first. https://t.co/id1VSv8mD9view source ↗
-
• twitter 1w negative@thdxr Staying on vanilla setups lets you absorb model improvements directly, whereas custom scaffolding usually just preserves workarounds for problems that no longer exist. @thdxr Staying on vanilla setups lets you absorb model improvements directly, whereas custom scaffolding usually just preserves workarounds for problems that no longer exist.view source ↗
-
• twitter 1w positiveSilicon Valley keeps building me a husband when what I really want is a wife lol I don't want an agent that only does one-off, random tasks that are nice but not life-changing: booking flights, making dinner reservations, or buying tickets I want an AI wife who knows what’s on the family calendar, remembers there’s a… Silicon Valley keeps building me a husband when what I really want is a wife lol I don't want an agent that only does one-off, random tasks that are nice but not life-changing: booking flights, making dinner reservations, or buying tickets I want an AI wife who knows what’s on the family calendar, remembers there’s a birthday party Saturday and we haven’t bought a gift then sends me options, realizes we’re almost out of diapers and buys them, knows which bills are coming up, remembers someone needs a dentist appointment and books it, figures out what the hell we’re eating all week and what we need from the grocery store, and generally keeps track of the 47 things that need to happen before anyone else even realizes they need to happen. So, who's building me an AI wife that carries the mental load instead of an AI husband that waits for me to give it a chore??view source ↗
-
• twitter 1w neutralHow do I take down brands running Meta ads with 100+ stolen creatives from us? Literally downloaded & re-uploaded with no changes. Or am I taking an L? How do I take down brands running Meta ads with 100+ stolen creatives from us? Literally downloaded & re-uploaded with no changes. Or am I taking an L?view source ↗
-
• twitter 1w neutral0 events, 0 dinners, 1k PRs merged for tech week gang 0 events, 0 dinners, 1k PRs merged for tech week gangview source ↗
-
• twitter 1w neutralwe recreated interfaces for the agentic era 🤯 no windows. no dashboards. every screen is built the moment you ask, then disappears. try now at https://t.co/x89j4hrZMY https://t.co/LHWTdSLUhG we recreated interfaces for the agentic era 🤯 no windows. no dashboards. every screen is built the moment you ask, then disappears. try now at https://t.co/x89j4hrZMY https://t.co/LHWTdSLUhGview source ↗
-
• twitter 1w positiveHouse call for the Falcons 🔥 Bijan Robinson takes it 59 yards to the house! https://t.co/iJ15xrURzC House call for the Falcons 🔥 Bijan Robinson takes it 59 yards to the house! https://t.co/iJ15xrURzCview source ↗
-
• twitter 1w neutralCoding agents don’t have to trap you at the terminal. Heard is an open-source voice layer for coding agents, built for developers who want spoken updates from their AI workflows. It helps you stay on top of agent work away from the screen by turning activity into narration and surfacing the moments that need attentio… Coding agents don’t have to trap you at the terminal. Heard is an open-source voice layer for coding agents, built for developers who want spoken updates from their AI workflows. It helps you stay on top of agent work away from the screen by turning activity into narration and surfacing the moments that need attention. Key features: • Three listening modes – choose Co-pilot, Companion, or Focus for short updates, fuller briefings, or alerts only • Agent integrations – supports Claude Code, Codex CLI, and Codex App; `heard run` can wrap other coding CLIs • Multi-agent awareness – voices the most salient blocked, decision, or failure event and summarizes the rest • Self-hosted path – build from source with your own voice/narration keys or fully local Kokoro • Personas and repo settings – includes four personas plus per-repo `.heard.yaml` tuning It’s open-source (Apache License 2.0). Link in the reply 👇view source ↗
-
• twitter 1w positiveLove it - great idea. Democrats will hate this. Love it - great idea. Democrats will hate this.view source ↗
-
• twitter 1w neutralThe Pentagon announced Monday that Nidal Hasan, who was sentenced to death for the 2009 mass shooting at Fort Hood, will be executed by firing squad, marking the first U.S. military execution in more than 60 years. https://t.co/dc4TgFCNnB The Pentagon announced Monday that Nidal Hasan, who was sentenced to death for the 2009 mass shooting at Fort Hood, will be executed by firing squad, marking the first U.S. military execution in more than 60 years. https://t.co/dc4TgFCNnBview source ↗
-
• twitter 1w neutralSimu Liu got married to his fiancée, Allison Hsu. [📸: @simuliu ] https://t.co/Filp7WHu9j Simu Liu got married to his fiancée, Allison Hsu. [📸: @simuliu ] https://t.co/Filp7WHu9jview source ↗