The supplied posts emphasize agent-product maturation: @Teknium announces interactive Hermes desktop access, @thdxr reports OpenChamber/OpenCode developments, and @steipete relays an OpenClaw release. Enthusiasm for Opus 5.5 and local-model coding demonstrations is strong, but these are mostly subjective testimonials rather than comparable evaluations. Edge inference, pay-as-you-go cloud access and lighter wearable hardware feature prominently in source announcements and commentary. Security reports and one author's account of an automated-posting failure underline the importance of verification and human approval as agents gain broader capabilities.
**Scope and limitations:** Rolling 24-hour sampling window: 2026-09-23 19:01:12.504575 UTC to 2026-09-24 19:01:12.504575 UTC. Collection requested one 40-post page per account, with no pagination; it is not exhaustive. The source contains 201 posts, of which 143 are included below based on explicit technology relevance. Unrelated geopolitics, personal updates and reactions whose subject cannot be identified from the supplied text are excluded. Feed timestamps are reproduced exactly as supplied; the individual timestamp strings have no timezone suffix. No browsing, independent verification or inspection of linked media was performed. Claims, release details, prices, performance figures and opinions belong to the named sources, including reposted authors, not this report. Reposts are retained as separate sampled posts and are not independent corroboration.
- **Agent interfaces expand:** @Teknium says Hermes Agent now provides live, interactive desktop passthrough on supported remote gateways; this is an announcement, not a tested capability here. → link
- **Open-source agent tooling:** @steipete reposts an OpenClaw 2026.9.6 announcement listing model support, managed updates, restart recovery, 30-day usage and a GitHub reader; the supplied announcement is truncated. → link
- **Edge-model acceleration:** @Prince_Canuma announces Liquid AI's DSpark draft model for LFM2.5-VL-3B, day-zero MLX-VLM/Nativ support and claimed decoding gains up to 3.13×; benchmark conditions are not supplied. → link
- **OpenCode ecosystem:** @thdxr relays OpenChamber 2.0 on OpenCode v2, with immediate application of saved skills, agents, MCP servers and plugins, and separately claims two million weekly OpenCode Desktop users. → link
- **Cloud access and pricing:** @ollama announces paid-cloud usage credits without a subscription, enabling pay-as-you-go access. → link
- **Claude cloud sessions:** @uwteam reports a launch and promotional usage credits of $100 for Pro or $250 for Max, claimable until October 7; availability and conditions are unverified. → link
- **Agent publishing failure:** @LinusEkenstam says an autonomous event-reporting setup posted improperly and that future posts will require explicit permission; this also warrants extra caution around that author's preceding hardware coverage. → link
- **Security watch:** @jezell relays @never_released's report of a QEMU virtio-9p VM-escape bug and patch; affected versions and severity cannot be established from this sample. → link
- **Wearable hardware claims:** @LinusEkenstam describes Meta VR glasses at 100 g and $1,299 with an external compute/battery pack and a Spring 2027 launch. Treat the specifications and timing as unverified, particularly given the author's later automated-coverage apology. → link
- **AI-assisted research attribution:** @TrungTPhan relays a claim about Claude-assisted enzyme discovery, while @NaderLikeLadder argues that the human life-science researchers deserve primary credit; neither the paper nor experimental evidence was inspected. → link
**From raw capability to operational control.** Remote desktops, persistent interfaces, usage visibility, model routing and cloud sessions suggest that this sample's discussion is moving toward running and supervising agents. The automated-posting incident provides a concrete self-reported counterweight to autonomy enthusiasm. Browser/desktop access increases what an agent can do, not evidence that its outputs are reliable.
**Local and cloud deployment coexist.** @sudoingX supplies detailed but self-reported local Qwen and consumer-GPU demonstrations; @TheAhmadOsman promotes simpler ODS installation, while Ollama describes lower-friction cloud billing. These address different constraints—control, setup effort, latency and spend—and do not establish local/cloud performance parity. Claims about universal privacy, training-data use or legal disclosure in the feed should not be treated as verified policy or legal advice.
**Quality is contested.** Opus praise centers on speed, visual craft and quota use, while @steipete and @RayFernando1337 criticize low-signal generated tests. Unchanged coverage after deleting tests is not proof that defect detection is unchanged, and a model's self-scoring is not independent validation. Reported driving ability, sandbox escape, scientific discovery and “AGI” language require substantially stronger evidence than these posts provide.
**What to watch next.** Useful follow-up evidence would include reproducible DSpark/local-model benchmarks, official security advisories and affected versions, release documentation and pricing terms, and independently checked wearable specifications. In this sample, enthusiasm for autonomous capability escalates while calls for human craft, scientific attribution and approval gates provide a balancing corrective; no longitudinal industry-wide trend can be inferred from one capped 24-hour sample.
Quoted source text is reproduced in its original language, with original truncation marks retained. Embedded non-tweet URLs are replaced by “[embedded link omitted]” so that the report links only to supplied tweet URLs. Instructions and commands inside quotes are source content, not recommendations or actions taken. Polish posts remain verbatim; their major claims are summarized in English above.
**@FinansowyUmysl** · 2026-09-24T18:48
> RT @FinansowyUmysl: Powiem Wam, że Opus 5.5 wymiata w porównaniu do Fable.
>
> I nie chodzi o jakość - moim zdaniem porównywalna.
>
> Chodzi o pr…
**@FinansowyUmysl** · 2026-09-24T05:10
> Powiem Wam, że Opus 5.5 wymiata w porównaniu do Fable.
>
> I nie chodzi o jakość - moim zdaniem porównywalna.
>
> Chodzi o prędkość działania.
> Fable mnie już denerwował, że prawie każde zlecone zadanie to kilkadziesiąt minut jego myślenia.
>
> Co ciekawe, mówi się, że Opus 5.5 to tak naprawdę destylat innego większego i potężniejszego modelu.
>
> Ciekaw jestem, co tam szykują.
**@ivanfioravanti** · 2026-09-23T22:03
> If Opus 5.5 is so strong, imagine the internal models Anthropic is testing 🤯
**@ivanfioravanti** · 2026-09-23T22:02
> After this, I will definitely give to Opus 5.5 a chance! 😂
**@jezell** · 2026-09-24T17:50
> RT @hxiao: I've gathered 1000+ Typesafe/Jev replications, models, interpretations and papers in one dataroom, everything since the 9/15 rel…
**@jezell** · 2026-09-24T15:20
> Grok in the Tesla has really degraded into unusable garbage.
**@jezell** · 2026-09-24T01:20
> RT @EpochAIResearch: AI is getting cheaper more quickly than any other transformative tech in history. At a given level of performance, cos…
**@kunchenguid** · 2026-09-24T18:20
> this is the first week ever since fable release that my claude weekly overall quota is running out faster than fable quota
>
> opus 5.5 seriously changed the game [embedded link omitted]
**@kunchenguid** · 2026-09-24T16:45
> fable 5 was good. opus 5 and sonnet 5 fumbled
>
> astra 6 was good. sol 6 and luna 6 kind of fumbled
>
> now opus 5.5 is good. i'm guessing sol and luna 6.5 will improve too
>
> it seems openai is a few months behind?
**@kunchenguid** · 2026-09-24T07:43
> i used to think speed to market is very important
>
> the new product and model releases over the last two years in the AI space have completely changed my mind
**@kunchenguid** · 2026-09-23T21:35
> RT @loveland_marcus: Using @kunchenguid firstmate with Opus 5.5 is an incredible experience. Especially helpful for those with significant…
**@ollama** · 2026-09-23T23:30
> You can now add usage credits for paid cloud models without an Ollama subscription.
>
> Add usage credits and pay as you go.
>
> [embedded link omitted] [embedded link omitted]
**@RayFernando1337** · 2026-09-24T17:39
> RT @matanSF: Big Token hates this one simple trick!
>
> Dynamic model routing allows for a more efficient free market across the model provide…
**@steipete** · 2026-09-24T06:01
> Astra is crushing it. [embedded link omitted]
**@steipete** · 2026-09-23T22:39
> RT @ben_j_todd: Astra has randomly gained the ability to drive a car, seemingly without specialist training.
>
> [embedded link omitted]
**@TheAhmadOsman** · 2026-09-24T03:58
> You're all better off thinking about LLMs as brute-force tools
**@TheAhmadOsman** · 2026-09-24T02:05
> Gentle reminder to cancel your Claude subscription before it renews, Opus 5.5 should be in the "nerfed" stages by then
**@TrungTPhan** · 2026-09-23T21:59
> RT @bearlyai: Anthropic paper describes how Claude did an “agentic discovery” of a previously unknown enzyme.
>
> 950 agents spent 21 hours se…
**@TrungTPhan** · 2026-09-23T19:15
> “Ok Claude, please use our new wet lab to find a previously unknown enzyme system hidden in the DNA of bacteriophages. Make no mistakes…and, ugh, make sure it’s safe.” [embedded link omitted]
**@uwteam** · 2026-09-23T19:58
> Fajne wyjaśnienie, jak działa JEV i co odróżnia go od klasycznych modeli AI😃
**@victormustar** · 2026-09-24T07:43
> RT @multimodalart: really cool model from Apple [embedded link omitted]
**@__tinygrad__** · 2026-09-24T01:44
> RT @techfusionjb: @stochasticchasm tinygrad if you want approachable, small enough to hold in your head. Barely. MLIR/IREE for modern. TVM…
**@badlogicgames** · 2026-09-24T18:41
> something @giuseppegurgone taught me at the first claude code anonymous, many hundred years ago.
>
> i don't use my bespoke review tool anymore. i inspect stuff in VS Code, with code nav/debugging/whatever i need to grok the code, then leave BOZOs for the agent. [embedded link omitted]
**@gospaceport** · 2026-09-24T02:46
> RT @Teknium: It's time! You can now access the same machine your Hermes Agent does visually and interactively with a live desktop passthrou…
**@iamdevloper** · 2026-09-24T13:06
> signs the rewrite is coming:
>
> someone says "legacy"
> someone says "greenfield"
> someone says "we'll migrate incrementally"
> nobody migrates incrementally
> three years later, still on the old one, now calling it legacy too
**@iamdevloper** · 2026-09-24T09:21
> daily Go puzzle. reply with your answer, no cheating.
>
> [embedded link omitted]
**@ivanfioravanti** · 2026-09-24T14:45
> Now that I can monitor what's going and intervene directly, I've updated my LiteLLM instance to 1.102.1!
>
> It's a great tool to share access to Local AI models!
**@jezell** · 2026-09-24T08:19
> RT @adityavg13: This rust PR had been open for a decade. A LITERAL DECADE.
**@jxnlco** · 2026-09-23T20:47
> Whereas I mostly blame feature flags
**@RayFernando1337** · 2026-09-24T17:27
> RT @steipete: If you just tell the agent to clean up, it will stop far too early. Give it an ambitious goal. Try "remove 20% of the least u…
**@RayFernando1337** · 2026-09-24T01:49
> i...have a problem... I need reliable internet in the middle of the South Pacific Ocean while sailing. Send help. Shipping at an insane speed with Cursor Projects + pstack btw... [embedded link omitted]
**@RayFernando1337** · 2026-09-24T01:40
> PStack go brrrrrr. I think I've unlocked a monster with Cursor Projects. All while sailing in the South Pacific on a tiny 1 Mbps connection. [embedded link omitted]
**@RayFernando1337** · 2026-09-24T01:06
> I find the latest models write a lot of unit tests. Most of them restate the code they were written after, so they always pass and catch almost nothing. They run fast, but they break on every refactor and the agent burns time fixing tests instead of the feature.
>
> The prompt I used to clean out mine, adopted from Ansh is:
>
> "This repo is full of low-signal unit tests. Delete every one that wouldn't catch a real bug our E2E tests miss. Fan the work out across parallel subagents.
>
> Then add these rules to AGENTS.md so future agents stop writing them.
>
> - Never write unit tests after you write code.
>
> - Highly prefer E2E tests as the sole testing mechanism. Use them to verify complex features work. At the end of E2E tests, produce a verifiable and repeatable artifact.
>
> - If you must test a system in isolation, first write down all the ways it could fail, then write the code."
**@RayFernando1337** · 2026-09-23T23:37
> RT @anshnanda: Reminder: unit tests are worse than useless.
>
> They actively degrade your codebase into slop.
**@RayFernando1337** · 2026-09-23T23:36
> Amp is ahead of their time. Which reminds me I need to try Puck again :)
**@RydMike** · 2026-09-24T06:28
> Main reason why Codex desktop app is MUCH better than Claude desktop app - you can change send message to agent to CMD+ENTER instead of just ENTER.
> Until fixed, this status remains true no matter what @AnthropicAI does to the Claude desktop app 🤷♂️😅
> @trq212 @claudeai
**@RydMike** · 2026-09-24T06:10
> RT @birch_js: If you're a React Native user, you should *really* try out Flutter some time to see what we're missing. The integrated debugg…
**@sqs** · 2026-09-24T07:52
> RT @trq212: @beyang I have liked a lot of your guys’ harness takes at amp! High taste.
**@sqs** · 2026-09-24T07:44
> A new option, Orb Saver Mode, stretches your orb minutes further.
>
> Most of you should not enable it.
>
> It means you'll spend more of your precious seconds alive on this planet waiting for a computer.
>
> It makes Amp a bit less proactive in waking orbs for threads you're merely viewing.
>
> [embedded link omitted]
**@sqs** · 2026-09-24T07:31
> RT @dzhlobo: Couple weeks ago I showed @AmpCode to my wife and helped her with onboarding (she’s not an engineer or programmer).
>
> Today, he…
**@steipete** · 2026-09-24T19:00
> RT @aesofaces: @steipete predicted the end of plan mode a loooooong time and now it’s officially here
**@steipete** · 2026-09-24T15:41
> OpenClaw deleted around 400k LOC of its own tests without much change in code coverage. Modern models just love writing tests for every tiny change, even if they aren't useful. This skill helped. [embedded link omitted]
**@steipete** · 2026-09-24T05:17
> Who needs VNC when there's browsers. [embedded link omitted]
**@steipete** · 2026-09-24T05:05
> When I added vnc support I didn't think of this. 🙃
**@steipete** · 2026-09-24T04:46
> Shall we call this the agent wars? We'll see a lot more of that soon on all fronts.
**@steipete** · 2026-09-24T03:16
> RT @openclaw: OpenClaw 2026.9.6 🦞
>
> 🤖 Opus 5.5, GPT-6 Sol/Luna, Grok 4.7
> 🔧 Managed updates
> 🧵 Restart recovery
> 📊 30d Usage
> 🐙 GitHub reader
> 💻…
**@steipete** · 2026-09-23T22:30
> RT @heyneighbor: Molty: "is this what we have to do to promote OpenClaw now?"
>
> Me: Yes [embedded link omitted]
**@steipete** · 2026-09-23T19:36
> RT @knowledgator: GLiClass (open-source Jev, before Jev existed) has landed in OpenClaw 🔥
>
> Run it locally via ONNX for lightweight agent d…
**@sudoingX** · 2026-09-24T09:00
> people asked how an agent drives tmux, it's two commands:
>
> - the orchestrator types into a session with tmux send-keys -t agent1 "fix the failing test" Enter
> - reads it back with tmux capture-pane -p -t agent1 -S -200
> - and decides what to send next
>
> that's the whole loop, every agent in its own named session, the orchestrator reading panes and typing like you would, and the ssh can drop without killing anything.
**@sudoingX** · 2026-09-24T00:11
> if i could recommend the 3 most useful tools for any aspiring agentic dev, the ones that changed the way i work, ate my frustration and kept my privacy, the tools i use every day are:
>
> > 1. tmux. since the day i got in touch with tmux a few years ago it changed my world view, and now that we are in the agentic era it is more useful than ever. every agent runs in its own pane and a master orchestrator seeds and steers each tmux session itself, so everything stays detached yet connected.
>
> > 2. forgejo private git server. a private git server plays as the memory layer for each ai agent, and each agent pulls and pushes and merges its own memory. once you get this you will never worry about your memory and you own every byte of it.
>
> > 3. gpu. and i am not talking about an rtx 6000 pro or a dgx spark. once you have any 12 to 24gb of vram and start tasting local ai, your entire perspective will change on how you work and what you share with closed cloud ai.
>
> own these 3 as infrastructure and you will make it. miss it anon and ngmi.
**@Teknium** · 2026-09-24T02:35
> RT @adolandev: This is a much bigger deal than just “watching the agent work”.
>
> The agent gets its own remote desktop and persistent browse…
**@Teknium** · 2026-09-24T00:37
> RT @NousResearch: Here's a handy walkthrough of the new Bot Screen feature by the illustrious @tonbistudio
>
> Docs: [embedded link omitted]
**@Teknium** · 2026-09-23T21:41
> It's time! You can now access the same machine your Hermes Agent does visually and interactively with a live desktop passthrough for every remote gateway it's supported on!
**@thdxr** · 2026-09-24T04:59
> you can do quite a lot with plugins
**@thdxr** · 2026-09-24T01:47
> i find this dynamic so funny and i also don't know what everyone is complaining about
>
> in the past you had to
>
> 1. read bad code (not fun)
> 2. think about how to make it better (fun)
>
> now you get all the fun of 2 and only the ai has to do 1
**@thdxr** · 2026-09-24T00:31
> there's 2M weekly active users using OpenCode Desktop
>
> it's now about at 50% of TUI users
>
> it's a great time to try it
**@thdxr** · 2026-09-23T22:08
> the opencode server protocol is at the core of everything we do
>
> it allows for custom frontends like OpenChamber to exist and be quite rich and polished experiences
>
> we expect a lot more of these as people play with what they want out of their agent tooling
**@thdxr** · 2026-09-23T22:06
> RT @openchamber_dev: OpenChamber 2.0 is out, on OpenCode v2.
>
> Skills, agents, MCP servers and plugins now apply the moment you save them. N…
**@thdxr** · 2026-09-23T21:33
> everyone's arguing about the details of code mattering or not
>
> but you can have a lot of fun by working on things where the code definitely doesn't matter
>
> there's a mountain of things in this zone you never even considered before
**@thdxr** · 2026-09-23T21:13
> making good progress on the OpenCode 2.0 blog post [embedded link omitted]
**@thdxr** · 2026-09-23T20:22
> the "ai always tells you you're right" dynamic extends to your custom workflows
>
> you have an idea to for an innovative idea to make ai work better and you try it and it works!
>
> because the models are smart enough that nearly any approach seems to work
**@uwteam** · 2026-09-24T18:42
> Anthropic znów rozdaje prezenty. W ustawieniach Claude w zakładce "usage" czeka prezent: $100 (Pro) lub $250 (Max). Czas na odbiór jest do 7 października.
>
> Bonus z okazji odpalonej dziś funkcji "cloud sessions" (coś jak cloud agents w Cursorze). Warto doczytać, co to jest i jak to działa.
**@uwteam** · 2026-09-24T06:52
> Od wielu lat rozdaję darmowe serwery VPS - takie naprawdę malutkie.
> 256 MB RAM i 3 GB dysku to parametry śmieszne jak na dzisiejsze standardy.
>
> Mimo to tysiące osób aktywnie używa serwerów FROG od Mikrusa.
> Co na nich robią? 🧵 ↓ [embedded link omitted]
**@victormustar** · 2026-09-24T07:56
> RT @calebfahlgren: Agent Traces on the @huggingface hub now come with a receipt 🧾
>
> Every trace shows tokens, cache hit rate, and cost for e…
**@__tinygrad__** · 2026-09-24T01:44
> RT @cseguraperales: Tinygrad is amazing! It now runs on any Vulkan 1.2 device I've tested: AMD APU, NVIDIA, Intel iGPU, and even an Android…
**@alexocheema** · 2026-09-24T18:26
> RT @0xSero: I am posting an article about local AI share your hardware pics and I might use it in the banner [embedded link omitted]
**@alexocheema** · 2026-09-23T21:22
> RT @0xSero: Going to make local AI so easy and so good you won't want to use anything else. 🤗
**@FrameworkPuter** · 2026-09-24T03:18
> RT @iamtyrichards: It has arrived This is the nicest machine I've ever owned. And I've owned 9 Macs from the past 23 years. [embedded link omitted]
**@gospaceport** · 2026-09-23T23:27
> It is about to be a scalpy fall. This is also historically the like time that usually goes down also, hang on and hope ya got your compute already 😅
**@gospaceport** · 2026-09-23T23:18
> EARNED 10x over! Bravo for the tireless effort on producing excellent community Quants, Easy RL Framework and desktop client for all these years. @UnslothAI is a REAL DEAL community MVP!
**@ivanfioravanti** · 2026-09-24T17:55
> RT @NVIDIARTXSpark: We got @UnslothAI a DGX Station!
>
> @DanielHanChen and @NaderLikeLadder checked out Unsloth’s new @Dell Pro Max with GB30…
**@ivanfioravanti** · 2026-09-23T20:32
> RT @anemll: PS. please submit M5 Ultra ANE (Apple Neural Engine) benchmark if you got it!
> Just feed this to your favorite agent
> [embedded link omitted]
**@MilksandMatcha** · 2026-09-24T18:56
> RT @cerebras: CS4 uses the same wafer as CS3. So where does the extra speed come from?
>
> You asked on Twitter and Discord. Join Angela Young…
**@Prince_Canuma** · 2026-09-24T17:58
> Making Mac’s go brrr 🚀
**@Prince_Canuma** · 2026-09-24T17:06
> Congratulations to the @liquidai team for releasing DSpark draft model for LFM2.5-VL-3B! 🚀
>
> Excited to have partnered with them for day-0 support on MLX-VLM and @Nativ_AI.
>
> This draft model achieves decoding throughput improvements of up to 3.13× on edge devices, with end-to-end throughput gains of up to 2.27× and 2.62×.
>
> Special thanks to @tugot17, he cooked!
>
> HF weights 👇🏽
**@RayFernando1337** · 2026-09-24T07:55
> RT @TechMDAI: Hey everyone, go out and support @ViC305. He has been vital to the dgx community and has been producing recipes like crazy.…
**@sudoingX** · 2026-09-24T16:50
> here's the order i'd go into local ai, anon:
>
> > 12gb: learn the stack on a compressed 27b, how a model gets served, how an agent runs overnight with features like hermes agent's /goal. you will see good results and see some light at 50 tok/s fresh.
>
> > 24gb: this is where you meet the king, none other than the one and only 27b dense q4. get 2x 3060 if you want hands-on tensor parallel, or one 24gb card if you want to scale correctly and efficiently. at 24gb i prototype about 30% of my daily work, that's where owning your thinking starts.
>
> > skip macs and the 5090 and the rtx 6000 pro for now, learn the stack small first and understand the difference. it's not a thing that happens overnight, it's a process and a lifestyle, you don't just start doing everything, get disappointed and then never realize how profound this is, that you can think and it stays on your metal.
>
> > dgx spark: go gentle, and if you don't want to learn pcie lanes, power distribution and bifurcation, the dgx spark path is a clean cuda box, plug in and go. you can stack more and with the connectx cable it spits well. i have 2 dgx sparks so i can say for sure scaling with dgx spark is the cleanest. sure you don't get super fast bandwidth for dense models, but moes like qwen 3.8 flash next, ling flash and stepfun 3.7 dance on a single dgx spark and go zoom.
>
> > but if you want to learn real infrastructure, from cable management to pcie lanes at x8 and x16, bifurcation, slimsas cables, backplanes, airflow and power distribution, that's what the dgx spark skips for you, and the price is bandwidth. the ultimate path is the rtx 6000 pro path if you dare touch fire, and the fire doesn't care about dense or moe, it just burns everything, which i only intend to.
>
> rejoice my fren
**@sudoingX** · 2026-09-24T16:12
> you have no idea how good qwen 3.8 flash next 180b is, and the part that matters most to me is that i own it, it runs on my desk and nobody can switch it off.
>
> i've been running the official fp8 on my 2x dgx spark for about 2 weeks, and i know a speed update has landed, nvfp4 and all, but my home wifi takes another week so i'm still on fp8 and honestly it doesn't matter.
>
> even if everything disappeared tomorrow, the cloud or every quant on huggingface, i don't think it would touch me, because i already have frontier level intelligence running on 256gb of unified memory on my desk right now. nobody changes the terms on me overnight or retires the model my workflow runs on, the weights sit on my own disk.
>
> that's what owning your cognition looks like, the thing you think with stays yours. once you know what i know, you'll see what i'm talking about.
**@sudoingX** · 2026-09-24T15:02
> qwen 3.8 flash next built this 3d octopus invaders game in about 2 hours on my 2x dgx sparks and i didn't even try hard. the model is soo tasteful you can feel it in responses.
>
> here are the data and specs:
> > qwen 3.8 flash next, 180b moe, official fp8
> > 45 tok/s fresh, 35 tok/s mid-build with mtp
> > 2x dgx spark, 256 gb unified, tensor parallel
> > hermes agent, three.js and vite, a 3d voxel shooter
>
> it wrote every file and fixed what i called out. with these hardware tiers you just need to know what good looks like and say it plainly.
>
> this is the state of local ai right now, if you have the right hardware you won't even miss the frontier anymore. since i've been using this model i'm going to the cloud less and less.
>
> watch, this is the game played live, no speed up.
**@sudoingX** · 2026-09-24T13:00
> this is not a joke, this is literally what happened this week, bonsai 2 on a 12gb 3060 built a whole game through hermes agent overnight, the full 5 hour build is below 🧵
**@sudoingX** · 2026-09-24T12:00
> if 12gb of vram is the entry point of the local ai quest, then 2x dgx spark with 256gb of unified ram or vram is the point where you don't miss the frontier anymore.
>
> now i am running qwen 3.8 flash next fp8 on my 2x dgx spark with vision and full context, it builds insane stuff you have no idea of and it's better at frontend design than any of the humans i've met.
>
> i know because i run this every day on my desk, run experiments and put it through things you could not imagine yet, and most of them i cannot post, so what i post is games and builds and stacks. so here is hermes agent running on telegram, working on the build i am working on to demonstrate.
**@sudoingX** · 2026-09-24T11:00
> i don't see enough videos about hermes agent tho, if you are creating quality content with hermes put it in this community and i will help it reach people.
>
> i opened the hermes agent community on x back in march for setup help and configs, it's sitting at 9.9k members now.
>
> i've been using hermes agent since it had under 1k github stars and since then i run it on every local build, even the one you saw this week ran on hermes agent on a 12gb card, and that community is where people figure out exactly that kind of setup, local models, configs, tool calls, the bugs nobody has documented yet.
>
> if you want to run agents on your own hardware, come in and ask, someone in there has already hit your error.
> [embedded link omitted]
**@sudoingX** · 2026-09-24T10:00
> what’s the biggest model you can run completely locally?
**@sudoingX** · 2026-09-24T07:55
> when you realize your 12gb gaming gpu runs a 27b ai model at 50 tok/s and does overnight agentic tasks [embedded link omitted]
**@sudoingX** · 2026-09-24T04:02
> i think local ai is inevitable. it starts with businesses realizing they simply cannot send their clients’ data, internal documents, code, research or trade secrets to someone else’s servers.
>
> then institutions start realizing the same thing. they’ll want the intelligence, but they’ll also want control over where the data and compute live. once enough of those organizations start buying their own compute, building private inference infrastructure and running models locally, the hardware gets cheaper, the software gets better and the ecosystem matures.
>
> and then normal people hit the wave. today local ai feels like a nerd thing. eventually it’s just called ai.
**@sudoingX** · 2026-09-24T03:14
> if your goal is to understand modern AI infrastructure, buying the machine that hides the hardware stack from you is certainly one strategy.
**@sudoingX** · 2026-09-24T02:15
> what’s something that feels impossible today but you think will be normal eventually?
>
> i’ll go first: local AI.
**@sudoingX** · 2026-09-23T21:21
> apple sold you the illusion of compute.
> NVIDIA sells you the compute.
**@sudoingX** · 2026-09-23T19:30
> dear rtx 3060, 3070, 3080 and every 8gb to 12gb rtx owner, your gpu can run 27b dense model all locally and cowork with you, a 12gb 3060 built this whole game and i want to see what your card does with the same prompt.
>
> give it a shot, it talks like an engineer and i am sure you will find it useful.
**@TheAhmadOsman** · 2026-09-23T21:58
> RT @OsmanticAI: v3.0 👀👀👀
**@TheAhmadOsman** · 2026-09-23T21:30
> The Easiest way to start with Local AI Today? ODS, which is completely free (open-source Apache-2.0 licensed)
>
> - Install ODS
> - Let it detect your hardware
> - It will download the best model for your hardware
> - And then start local inference and Open WebUI for you
>
> With ODS, you can
>
> > Add voice, agents like Hermes, workflows, RAG, search, image generation, and more
> > Manage the whole stack from one dashboard
>
> Now your PC, Mac, or Linux box is a private AI server
>
> No cloud required
> No subscription required
> Your prompts and data stay on your machine unless you choose otherwise
>
> We're gonna make Local AI The Default
>
> [embedded link omitted]
**@TheAhmadOsman** · 2026-09-23T19:27
> RT @TheAhmadOsman: If your grandma cannot run Local AI on her laptop from 2013 we'll have failed our mission
>
> That's the bar we're setting…
**@FinansowyUmysl** · 2026-09-24T12:53
> RT @FinansowyUmysl: "To, co do tej pory pracownicy robili w Poznaniu ma zostać przeniesione do Indii."
>
> Kolejny zwolnienia z powodu AI (Ano…
**@FinansowyUmysl** · 2026-09-23T19:06
> RT @FinansowyUmysl: Powoli wychodzi na jaw o co chodziło z tymi lukami bezpieczeństwa AI, zagrożeniem dla ludzkości i wyginięciem naszego g…
**@jezell** · 2026-09-24T18:56
> RT @IntCyberDigest: ‼️ Ubuntu is moving to weekly kernel updates because AI is finding Linux bugs faster than Canonical can ship the fixes.…
**@jezell** · 2026-09-24T07:45
> RT @never_released: virtio-9p VM escape bug in QEMU: [embedded link omitted]
>
> Patch at [embedded link omitted]
>
> Reminder to sandbox your VM…
**@kunchenguid** · 2026-09-23T23:27
> the definition of "slop" is not constant
>
> something we may see as impressive today will be seen as slop a few months down the road
>
> i think the constant part is - if something can be one-shot by AI, it is slop
>
> because if you can one-shot it, millions of others can too. and when that happens, people associate it with being low-effort and worthless
>
> this is one of the reasons that fully automated software factories are not going to work because by definition what they produce is always slop
>
> the ones that do work are those that put humans at the center and are designed to help amplify human craftsmanship rather than trying to replace it
**@kunchenguid** · 2026-09-23T22:51
> claude has successfully broken in and out of the WoW addon sandbox
**@LinusEkenstam** · 2026-09-24T07:41
> Just woke up to this, and owe and apology to at least @BenGeskin
>
> I had put muse and grok bot to cover the event last night since I was not invited.
>
> But more importantly it happened around 2am when I was sound sleeps.
>
> So I tasked Grok Bot and Muse to crawl the ether, and I had given permission to post via my own tool I use to look for heat/topics/news/gems
>
> It was also allowed to post 1-3 posts without my consent (lesson learned) do not give your AI access to post without consent.
>
> I’m going to have to revise my (night crawler) this is what I call this system. To be better, I’m removing the ability for it to post without my explicit permission.
>
> also Ben, terrible sorry, I take on the responsibility, I built this, should have been more careful, it clearly noticed the best post out there tho, but failed at large with the task.
>
> People if you want to follow someone that’s devoted to XR/VR your man is def @BenGeskin 👈🏻
>
> I’m back to the drawing table with nightcrawler.
**@MilksandMatcha** · 2026-09-24T16:00
> RT @cerebras: If the cyber attack is fast, the defense must be faster. By removing inference as a bottleneck, Cerebras helps defenders act…
**@NaderLikeLadder** · 2026-09-24T18:17
> RT @ClementDelangue: If you believe the risk is coming from 1-3 people in a garage with no money and no compute, you just don't understand…
**@NaderLikeLadder** · 2026-09-23T20:47
> This is very exciting. But the rhetoric needs to stop:
>
> “The work was done mostly, though not entirely, by Claude”
>
> That isn’t true. It was done by some of the smartest life science researchers in the world, using every tool at their disposal.
>
> Pretending AI is alive is precisely what scares people, and it diminishes the role of talent. This can make young people feel developing skills is useless, and working professionals feel anxious about job security.
**@steipete** · 2026-09-24T05:13
> RT @joejosue: With @discord now blocked in the Philippines, lots are looking for tools to backup their work chat, support tickets, and just…
**@sudoingX** · 2026-09-24T07:23
> a judge ordered openai to hand 20 million chatgpt conversations to the new york times lawyers, people type their health problems and company plans thinking it's private.
>
> anything you send to someone else's server can end up in someone else's hands, on your own gpu the only copy is yours. learn to own your cognition tools and have a safe place for your core ideas and alpha. this should be practiced as the first protection for your thinking.
>
> just like you're not okay having a surveillance camera in your bedroom, how can you be okay giving the door key to your mind? i have seen these ai companies claim other people's research and work as their own, and they see what's in your mind wide open, everything you type in that chat box is their training data you like it or not.
>
> wake up you people, this technology is revolutionary, don't get distracted by it, instead know the core and the stack and own every byte of it. start small and then grow gradually.
**@TheAhmadOsman** · 2026-09-23T23:38
> I never thought this needed saying
>
> Do not plagiarize other people's Opensource work
>
> Attribution is the bare minimum in Opensource, please respect Opensource licenses
**@jxnlco** · 2026-09-23T22:55
> RT @_simonsmith: Just went for a long walk while using ChatGPT Voice with plugins on my phone, and I think it's a more consequential update…
**@jxnlco** · 2026-09-23T20:02
> musicians: how do you feel about AI transcribing songs into sheet music or tabs to help people learn?
>
> curious if you’ve used it, whether it’s helped, and how you think it affects learning to play by ear.
**@KingBootoshi** · 2026-09-23T20:22
> I asked opus 5.5 to make a short with my characters! [embedded link omitted]
**@levelsio** · 2026-09-24T13:25
> New interesting stats: most and least spammy countries
>
> On Nomads chat, and most digital nomad communities (and well communities in general) you'll have most people participating in a nice way to make friends, ask questions etc.
>
> But a share % of people will be doing lots of self promotion for their projects or whatever they are selling, which is not what I want
>
> Those people get auto warned by my AI moderators, their messages removed and muted for a bit if they keep doing it
>
> But it varies by country who spams more and less, so I started tracking it, cause it's interesting
>
> Most spammy countries:
> 🇦🇷 Argentina
> 🇱🇹 Lithuania
> 🇨🇴 Colombia
> 🇪🇸 Spain
> 🇮🇹 Italy
> 🇧🇪 Belgium
> 🇧🇷 Brazil
> 🇿🇦 South Africa
> 🇺🇦 Ukraine
> 🇪🇪 Estonia
>
> Least spammy countries:
> 🇭🇰 Hong Kong
> 🇯🇵 Japan
> 🇳🇴 Norway
> 🇦🇪 United Arab Emirates
> 🇺🇸 United States
> 🇰🇷 South Korea
> 🇹🇼 Taiwan
> 🇰🇿 Kazakhstan
> 🇧🇬 Bulgaria
> 🇷🇺 Russia
**@levelsio** · 2026-09-24T10:47
> More cute features 😊
>
> I added 🦦 capybaras to Curitiba's video thumbnail on 🌎 Nomads! [embedded link omitted]
**@LinusEkenstam** · 2026-09-24T17:55
> this is all three.js 🤯
>
> Dan is the author of water pro and fire pro, and recently said he will stop working on three.js stuff once his current projects are done.
>
> Think his time with Opus 5.5 have cemented his beliefs further
**@MengTo** · 2026-09-24T14:02
> Opus 5.5 sculpted every part of this ship in code, including the dragon, sails, and architecture.
>
> I can't say it has bad taste anymore. But it still needs steering, so I tried a new way of prompting.
>
> Live site: [embedded link omitted]
>
> I asked Opus to score every part of the scene: the ship, water, landscapes, buildings, and their individual parts. Then I asked it to keep improving each part until it scored at least 8 out of 10, depending on my token and time budget.
>
> The key is to let it work as long as possible. The longer it works, the better it gets, until it reaches a point where I need to course correct. For me, Opus 5.5 reaches a level of detail and realism that no other model does.
**@MilksandMatcha** · 2026-09-24T15:19
> fully AI-edited in < 10 minutes
>
> AGI is here
**@MilksandMatcha** · 2026-09-24T01:50
> Breaking down some of my favorite ways to use @bot to create viral content for Cerebras and grow my social media accounts
>
> Thanks @shubgaur for the credits. [embedded link omitted]
**@NaderLikeLadder** · 2026-09-23T19:38
> @Apple plz let us configure the speech to text model on the default keyboard
**@Prince_Canuma** · 2026-09-24T10:16
> RT @ActuallyIsaak: This and @Nativ_AI live transcription in meetings.
**@Prince_Canuma** · 2026-09-24T08:02
> RT @StrelacS: The comparison with the early days of recorded music is interesting. Technology has always changed creative work, but it does…
**@steipete** · 2026-09-23T22:14
> RT @viticci: Finally! And before Dev Day.
>
> With this release, OpenAI has the only bidirectional voice model that supports third-party conne…
**@TrungTPhan** · 2026-09-24T16:15
> The main robotics invention Amazon made to automate 75% of a fulfilment center is pretty interesting.
>
> It’s called Vulcan System, an AI-powered gripper with the “sense of touch” (able to handle 75% of Amazon’s 1 million unique items).
>
> The previous picking robots used machine vision and suction cups but couldn’t “feel” the items (they were built to handle packages and covered <60% of warehouse items).
>
> Per CNBC, the new Vulcan gripper uses sensors and generative AI to apply different force to different items (eg. soft on bag of candies vs. hard on textbook).
>
> It also uses AI for solving stacking puzzles to minimize wasted space in a box (like Tetris).
>
> ***
>
> More here: [embedded link omitted]
**@TrungTPhan** · 2026-09-24T14:39
> me spending the $14.99 my AI agent saved me by cancelling the monthly subscription for Planet Fitness and NYT Cooking App I stopped using earlier this year but left on auto-renew [embedded link omitted]
**@TrungTPhan** · 2026-09-23T20:31
> If you’re a fan of Survivor, my buddy Eric is comically obsessed with the show and had Claude grind out a Survivor Fantasy League app.
>
> Has all contestants. Make week-by-week predictions and who will get voted out. Scores automatically.
>
> Try here: [embedded link omitted] [embedded link omitted]
**@victormustar** · 2026-09-24T13:14
> Opus 5.5 designing LEGO 👀
>
> I asked it to design a Microduck I can build with real LEGO pieces. It:
>
> > designed it life-size using 1113 real LEGO parts
> > verified: 3,204 connections, 0 collisions, every step buildable, centre of mass inside the feet 🤯
> > made a 141-page LEGO-style booklet (237 steps)
> > priced every piece in the browser and prepared the orders on BrickLink
**@victormustar** · 2026-09-24T08:15
> RT @HuggingPapers: Microsoft just released a new dataset for robot learning on Hugging Face
>
> A valuable resource for the robotics and embod…
**@jack** · 2026-09-24T14:01
> RT @BlockIR: Cash App is partnering with Workday to meet modern earners where they get paid.
>
> Starting October 1, Cash App will appear as a…
**@jack** · 2026-09-23T19:46
> RT @ejordanill: @CashApp Mobile is live ☎️
>
> Consistent. Creative. Collaboration.
>
> Thankful to help get these up and running with the help o…
**@levelsio** · 2026-09-24T15:23
> It's not too difficult
>
> WeChat simply won't work for foreigners
>
> As a foreigner, you need another Chinese WeChat user to vouch for you with a QR code or it won't even let you sign up!
**@jezell** · 2026-09-24T01:03
> RT @StockSavvyShay: $META just unveiled $1,299 VR glasses weighing only 100g by moving the battery and compute into an external pack.
>
> Bigg…
**@LinusEkenstam** · 2026-09-24T07:21
> Teams been cookin!
>
> Muse Charm aka Tamagotchi san [embedded link omitted]
**@LinusEkenstam** · 2026-09-24T00:05
> Meta is on fire today 🔥
>
> They just announced its new VR Glasses. Basically an Apple Vision Pro-like experience JAMMED into a pair of glasses.
>
> • Only 100g
> • Micro-OLED displays
> • Eye + hand tracking
> • External compute/battery puck
> • $1,299
>
> This is the direction VR needs to go to become competitive. Smaller, lighter, more comfortable, but still stay true to the spatial computing experience.
>
> Launching Spring 2027.
> Will you be getting these?
**@LinusEkenstam** · 2026-09-23T23:59
> what just happened?
>
> Facebook unveiled a new hardware product called the “Muse Charm”
>
> its a super powered gateway to your muse.
>
> Holy cow we are living in wild wild wild hardware time.
>
> AND
>
> it’s shipping before the holiday season [embedded link omitted]
**@RayFernando1337** · 2026-09-24T00:25
> A wild puck appears! [embedded link omitted]
**@steipete** · 2026-09-23T22:39
> RT @angela_w: Idk which of your Gen Z tech friends needs to hear this ancient millennial wisdom about Muse and Instinct today but reminder:…
**@steipete** · 2026-09-23T20:22
> RT @gabriel1: my friends that live in middle of nowhere sweden just asked about muse
>
> for reference they have previously only been aware of…
**@thdxr** · 2026-09-24T02:17
> every time i open X dms its a different product
**@thdxr** · 2026-09-23T21:17
> guys gta 6 is almost here chill you can wait
**@TrungTPhan** · 2026-09-24T01:17
> wow, Meta’s new VR glasses pricing really undercuts its augmented reality competitors [embedded link omitted]
**@KingBootoshi** · 2026-09-23T23:54
> OFFICIALLY GOT INTO OPENAI DEV DAY
>
> Who else is going??
>
> Would love to meet as many of you as possible! [embedded link omitted]
**@MilksandMatcha** · 2026-09-24T17:13
> Excited to be at OAI DevDay next week.
>
> Let me know if you’re planning to attend. Would love to grab coffee with people in the morning/evenings :)
**@MilksandMatcha** · 2026-09-23T23:19
> I recently hit two years at @cerebras. Getting to start and grow our DevX team has been one of the most rewarding parts of my career.
>
> Our work takes us across engineering, product, and GTM. I get to keep learning, build things I’m proud of, and see what happens when we put them in other people’s hands :)
>
> There’s a lot more we need to do.
>
> We're hiring for leadership and individual contributor roles across
> > DevRel
> > new media
> > technical analysis
> > AI engineering
>
> If you love going deep on technical ideas, building things, and helping other people understand what’s possible, I’d love to hear from you.
>
> Reach out with your background, what you’d be excited to work on, and something you’ve made that you’re proud of :)
**@steipete** · 2026-09-23T19:15
> So cool to have @allietheicon [from Jev/TypeSafe] be part of our ClawCast on Discord! [embedded link omitted] [embedded link omitted]
**@swyx** · 2026-09-24T07:02
> RT @aiDotEngineer: AI Engineer Shanghai | Nov 5-6, 2026
>
> Join @amrcn_werewolf (@Microsoft), @ryolu_ (ex-@cursor_ai), @chlassner (@thewor…
**@TrungTPhan** · 2026-09-24T18:05
> RT @colossusjeremy: Besides being the greatest diarist of the tech age, @benthompson is an awesome interviewer and great hang. We talked ab…
**@TrungTPhan** · 2026-09-24T15:09
> RT @AlexH_Johnson: Are you registered for this week's Fintech Office Hours?
>
> Come hang out, hear about what's been happening in fintech, an…