@plotarmordev
Account based inUnited States!
About this account
- Account based in
- United States
- Connected via
- United States App Store
X says this location may be affected by a proxy or VPN.
Account-level information from X, not a live location or the device used for a specific post.
Cofounder & engineer. Running local AI, building things with it, and sharing what I learn.
Joined April 2022
- Tweets9.4K
- Following771
- Followers8.1K
- Likes16.2K
Pinned Tweet
This week, Drama TensorFold clones voices!
Give it a voice you have permission to use, and it speaks in that voice on your own GPU.
🔊 Meet Tomorrow, my assistant, running on it. Sound on!
github.com/plotarmordev/dram…
Really cool to see my Drama 3 repo as a dependency in this Jarvis project 🔥
Next up from me is a very high quality Drama 3 quant. It cuts the size in half and keeps all languages, so the voice takes up a lot less of the Spark/VRAM and allows to run more things alongside it!
LocalRouter AI assistant. One DGX Spark. Full duplex.
This actually feels like Jarvis in a box! Built on @ashxhart TensorFold
🧠 Qwen 27B
🗣️ Drama 3 with streaming (Thanks @plotarmordev )
🖼️ Qwen Image Turbo 2.1
🎬 MiniMax H3 video
👂 Whisper STT
~0.5s after I stop talking, it starts with its first word. Interrupt it anytime. Set the delay as you please for responses.
Normal models stay loaded and can be swapped as needed.
Want a video model up? Then Whisper and Drama will make space. Once the video is generated, start voice back up when the video is finished.
The goal is actually fully modular, so consider this a super early version of what it could be. MCP by default, but also ships with a simple UI you can extend as you please.
Thin by design, so everyone can make it their own.
What features do people want next? What about TP=2 or Mac Studio versions?
Finally, Kimi K3.1 looks close!
A model called Kibitz showed up on LMSYS Arena today and says it's Kimi. A screenshot from yesterday shows @Kimi_Moonshot Kimi Code tool already naming K3.1 in its model switch warning.
Last time, K3 sat on Arena as Kivine for about 30 hours before it launched in July. Nothing official from Moonshot yet, but if that pattern holds we're talking days.
Can a 163 MB speech model be a local WisprFlow? Bark 3 is as accurate as NVIDIA's Parakeet, which is about 15x bigger!
Similar thing I've been chasing with drama3, getting it to run on 20GB GPUs so more people can actually use it. This actually has potential to live anywhere.
Today we’re releasing Underdog Bark 3.0, the #1 frontier speech model
It's a tiny 163 MB download as accurate as NVIDIA’s Parakeet (15x bigger). It beats OpenAI's Whisper large-v3 (57x bigger)
It's fast- transcribes 1 min of audio in <0.25s on a Macbook. On recorded meetings (AMI), Bark 3.0 gets 7.82% of words wrong. Parakeet gets 9.42%. OpenAI's Whisper large-v3 gets 13.63% error rate
Bark runs on both your Laptop & Phone. It’s free and open source under CC-BY 4.0
Try private on-device meeting notes & dictation, powered by Bark, in Underdog today: underdog.ai/bark
netrunner retweeted
sparkDash 2.70 is out ⚡️
This is a MAJOR release.
New pages
- Cost compare: see what your fleet's token usage would cost on any hosted model, including cached-prefill savings
- Temps: daily and weekly GPU temperature graphs for every unit
- Metrics history per unit (48h and 30 days, opt-in), with a History panel on the unit page
Engines & hardware
- New engines: Strata & Flare (tok/s and benchmarks) and Radiance (live rates)
- Intel Arc GPU support
- AMD GPU telemetry
- GPU clock cap shown on each unit, and you can set it from Edit unit. If the unit has no passwordless sudo, the dashboard asks for the password, or you can run it on the Spark yourself
Benchmarks
- Decode bench: pick the temperature (default 0.7) and turn thinking on or off for prose. Both show on copied results and the share card
- Tool Eval Bench "Show card": a shareable image of a finished result
- Comparison card: tick two runs and get one image showing the difference
- Compare a run from one Spark with a run from another, plus an expandable scenario list
- MAX_ACTIVE_BENCH_JOBS setting to raise the fleet benchmark limit
Overview & monitoring
- Overview cards list every loaded vLLM model, plus CPU temp, CPU power and the engine's parallel slot count
- Token totals for the last 24h, 6h and 1h, with one row per unit and model
- Optional card showing inference request activity
- Alerts to ntfy, Discord, Slack or a webhook (opt-in), plus a tab badge and desktop notifications
Other
- "Run both" button in the Models panel
- "Back to Overview" link on every unit page
- Published container image
Link below.
We've entered a new era of AI where no console is safe anymore.
Somewhere along the way, maybe around GLM 5.3, everything started to change.
i jailbroke the nintendo switch 2
so i modded cinnamoroll, peter griffin, hatsune miku, and more into Mario Kart World, a switch 2 exclusive
now watch my friends and i get cooked by bots
learn more and get the exploit pepsipu.com/blog/2026-10-swi…
I looked into this and long context on Flint is almost free in memory!
Only 14 of its 49 layers keep KV for the whole sequence. The other 35 only look at the last 513 tokens. That works out to 28 KiB of cache per token, so 100K tokens of context fits in under 3GB.
Flint by @Recursive_SI released a model that at 4bit should come roughly at 20GB, so it sits comfortably on a 128/64GB Spark.
Recursive's own number is 150+ tok/s on a 64GB RTX Spark in NVFP4. Custom kernels account for 55% of that speed without touching the weights! Looking forward to test this along with drama on the same machine ~
Never name your agent after a time.
Mine was called Tomorrow, so everything it delegates shows up as "Tomorrow's message" and the other agents read it as "do this tomorrow" 😂
Fully automated procrastination…
I wrote a memory system for Tomorrow, the assistant you've heard me talk about, and it's the reason Tomorrow now works almost autonomously a lot of the time.
A few days in and it's going really well. What breaks for me is the interface. I've been running everything through Telegram with a single Hermes by @NousResearch, and managing multiple projects that way turned into a total mess. I thought it would be acceptable with some UX updates I added, but it's totally impossible.
I'm looking for something better. Now I’m considering mousebot or monocode. Anyone have experience with either?
Or anyone knows options that could give grok bot experience locally? I've also heard that Hermes is close to releasing a mobile app, would be nice if it has a @bot like system for agents but smh I doubt…
Flint by @Recursive_SI released a model that at 4bit should come roughly at 20GB, so it sits comfortably on a 128/64GB Spark.
Recursive's own number is 150+ tok/s on a 64GB RTX Spark in NVFP4. Custom kernels account for 55% of that speed without touching the weights! Looking forward to test this along with drama on the same machine ~
Introducing Flint, @Recursive_SI’s first model: a fast local coding model built with our automated research system, the Eureka Machine.
• Best-in-class fast local coding
• Up to 2x faster than comparable models: 125+ tokens/sec on a MacBook Pro, 150+ on RTX Spark
• Built from scratch in under 4 months
Private, unmetered intelligence on your own machine. Open weights available now on @huggingface, run it locally on your favorite framework or try it for free on @opencode or @pidotdev.
Up to 43% more tokens per second on MoE models, just from better kernels.
Those kernels are only one part of each step. The model also spends time picking which experts to use and running attention, so a speedup this big showing up in overall speed is huge!
Quantization is easy. Fast quantization kernels for every format × every GPU × every workload are not.
Today we're open-sourcing Humming with Ant Group: one kernel library for all of it — up to 5x kernel speedups, +14–43% E2E MoE throughput.
Blog + benchmarks: vllm.ai/blog/2026-10-21-humm…
Working on a very high quality quant of drama3 by @FishAudio that may cut its weight size in half 🚀
This week, Drama TensorFold clones voices!
Give it a voice you have permission to use, and it speaks in that voice on your own GPU.
🔊 Meet Tomorrow, my assistant, running on it. Sound on!
github.com/plotarmordev/dram…
I wrote a memory system for Tomorrow, the assistant you've heard me talk about, and it's the reason Tomorrow now works almost autonomously a lot of the time.
A few days in and it's going really well. What breaks for me is the interface. I've been running everything through Telegram with a single Hermes by @NousResearch, and managing multiple projects that way turned into a total mess. I thought it would be acceptable with some UX updates I added, but it's totally impossible.
I'm looking for something better. Now I’m considering mousebot or monocode. Anyone have experience with either?
Or anyone knows options that could give grok bot experience locally? I've also heard that Hermes is close to releasing a mobile app, would be nice if it has a @bot like system for agents but smh I doubt…
SECURITY WARNING: check your X sessions right now.
Mine says I have 4994 other active sessions. Nobody logs in that many times on their own devices.
If clicking a link on X really lets the site you land on grab your X session once that site gets hacked, this is exactly what it would look like.
Go to your sessions in settings and hit "Log out of all other sessions".
This baboon Elon made it so that if you click any link on here, that website you go to can get your twitter account info if it gets compromised
Which is how I have 3k+ sessions apparently
Readers added context they thought people might want to know
High session counts are a recent X app glitch from in-app browser use creating phantom Web sessions. External sites get no account info or access. Log out others via Settings > Security and account access > Apps and sessions.
help.x.com/en/managing-yo…
Of these three, the one that's real is also the funniest.
eMudhra, a certificate authority that browsers trust, had eight certificates whose secret keys were tiny numbers, all under 256. One of them was literally 4. Anyone could find them by trying 256 numbers, then use them to fake any website. Firefox and Chrome have blocked them since Oct 17, and Apple and Microsoft did the same by Oct 20.
The iPhone 11–16 one comes from a leaked Cellebrite flyer for an unlock tool that has to be within about 8 meters of the phone, and it targets older iOS versions.
Regardless of everything, times are changing
In the last 24 hours:
1. iPhone 11 through 16 are apparently vulnerable to remote attacks.
2. SSL certificates are broken until browsers remove eMudhra CA.
3. PHP >9 vulnerability CVE published (tomorrow).
The world is about to witness things that have never been seen before.
Readers added context they thought people might want to know
The post's iPhone claim is inaccurate; a leaked Cellebrite forensic tool targets a Bluetooth bug patched in iOS 26.5 and does not affect updated devices. The eMudhra CA and PHP issues are real incidents but framed sensationally.
justin.sh/blog/2026/10/c…
blog.mozilla.org/security/2026/…
php.net
CAREFUL: Your model's logs might say success while your agent reads half an answer!
SSE (Server Sent Events) keeps one HTTP response open and pushes the answer to you token by token as the model writes it.
When you stream with SSE, the server sends the HTTP status code, 200 OK, the "request succeeded" signal, before the model has finished decoding. So the request might still die halfway through, and your logs will show it as a success.
When a self hosted endpoint is feeding agents and you only check the status code, the agent might just keep working off that half answer. Check that the stream actually ended cleanly before treating the output as done.
My tool, TheGrill never trusts a 200. A stream that dies halfway counts as a failure, not as a fast answer: github.com/plotarmordev/theg…
"Lossless" abliteration means lossless on the questions checked.
Abliteration edits the weights to strip out refusals, but you’d need to figure out how well everything else holds up.
With the scores out in the open, you can see exactly what was covered and what’s left to check on your own workload, and that’s where and how you should test any model weight.
netrunner retweeted
I finally have some numbers behind my DeepSeek V4.1 Flash 1017 quant. EAQ v1 is 11% to 38% closer to the original than LibertAIDAI's EXL3.
Seems like the gap is bigger on longer context. To be clear, LibertAIDAI's quant is NOT bad. It's a solid quant, and it gave me a strong baseline to beat.
Already in the works: EAQ v1.1, v1.2 and finally EAQ v2.
Full details in the model card:
huggingface.co/Mia-AiLab/Dee…
