News

The bell is no longer the only place a notification shows up. Settings → Telegram notifications: save your Telegram username, press Link my Telegram, press Start in the bot — from then on every notification lands in both places at once: the web UI keeps the record, Telegram gets a copy.

• Same events, same moment — support replies, GPU alerts, a command waiting for you to sign in. One writer, two destinations.
• Per user, private — each account gets its own chat; nothing about your notifications is visible to anyone else.
• On or off whenever you want — one switch on the same page, or send /stop to the bot.
• The web UI never depends on it — Telegram is a mirror. If the bot is slow or down, the notification is still in the bell.

One thing worth knowing: Telegram does not let a bot message an account that has not started it, so the username alone cannot deliver anything — that Start press is the step that binds your chat. The page says so instead of pretending otherwise.

Open Settings, add your Telegram, and stop checking the site for a reply.

I was forced to move all VPS systems to another platform. The servers are being rebuilt and every machine is backed up first — details below.

Bad news: several serious problems with the Kali VPS service came to light.

• Hidden commission. The provider adds a markup that is not visible at first glance.
• Ban risk. A few complaints put the account under threat — and a ban takes every user's VPS with it.
• Quota. Without KYC I can only rent a limited number of VPS.

So every account now runs its own proxy pod, with automatic rotating backups in case an account is banned, separate payment methods for each — and a few more headaches.

That costs money, so prices for the hosted VPS go up significantly. In return, there is a new feature: you can connect your own VPS — free.

In the next 24 hours all current pods move to a different account: every old VPS is backed up, then destroyed. You can then rent a new one at the updated price and restore it from the backup (you can permanently delete your backup at any time or disable it), or install that backup on your own VPS with a password or private key.

Open Kali VPS to see the current prices.

DeepSeek V4.1 Flash Abliterated and V4 Pro Uncensored are on the LLMs page. No GPU to start, no session to stop — you pay per token.

• Flash — $0.53 in / $0.84 out per 1M, cached tokens at $0.12. The cheapest model on the page.
• Pro — $0.88 in / $1.40 out per 1M when the work needs more thinking.
• 1M context on both, always available — nothing to start, nothing to wait for.
• Fast tier of Flash at $2.45 / $3.92 for when latency matters; ~4.7x the price, so only for that.
• Served by a third-party API that may log conversations — no sensitive data there.

Good for command runs and bulk work, where GLM-5.3-abliterated — my pick for hard reasoning — gets expensive.

Open LLMs and pick a DeepSeek model.

FLUX.3 klein is live after Qwen-Image 2.1, and Drama 3 after AuK — both on Media.

• FLUX.3 klein — 5.6B model for text → image, reference editing and bounding-box layout. The prompt upsampler is built in.
• One card. int8 on 24 GB, bf16 on 48 GB, up to 4K.
• Drama 3 — 35B speech model for multi-character scenes: up to 16 speakers, inline cues, voice cloning from a few seconds of audio.
• 24 GB GPU + 64 GB host RAM, because its experts live in system RAM.

Workflows: flux3klein-text-to-image, flux3klein-image-edit, flux3klein-multi-image, drama3-text-to-speech, drama3-voice-clone, common-audio-to-srt.

Open Media → FLUX.3 klein or Drama 3.

Qwen3-Coder-Next is no longer on the LLMs page. It was the most expensive model I ran — 159 GB of weights that needed either one very large Blackwell card or two 94 GB cards working together — and at $9.45 an hour for a single user it never did enough more than the 35B models to justify that price.

• Why it went. The coding work is covered. Ornith 1.5 35B is the agentic-coding workhorse, Qwen3.8 27B takes the harder reasoning, and Ornith 1.5 9B is the cheap pick for everyday code.
• What it cost. $9.45 an hour for a solo user — several times the 35B models, for the same kind of work.
• Already mid-task? Move the work to Ornith 1.5 35B — the closest replacement for coding agents.

Open LLMs to see the current lineup.

Qwen-Image 2.1 is on the Media page, at the top of the image list. One 7B model does what the previous Qwen bundle needed two for.

• Text → image. Renders at a native 2K (2048×2048), and the lettering on a sign or a poster comes out spelled right.
• Editing with up to 10 references. Upload the picture you want changed plus anything it should borrow from, then point at them in the prompt as <image1>, <image2> … Faces, products and the rest of the frame stay put.
• Transparent PNG, natively. Ask it to drop the background and you get a real alpha channel — no matting pass, no white halo along the hair line.

Three ready-made workflows ship with it: qwen21-text-to-image, qwen21-image-edit and qwen21-background-removal. Runs from 24 GB.

Open Media → Qwen-Image 2.1.

GLM-5.3-abliterated is in the model list: a frontier-class model with the refusals removed, and fully tested on my platform before release.

• Always ready, no GPU needed. There is nothing to start and no model to wait for. It answers straight away, and nothing is billed while it sits idle.
• Pay per token, not per hour. $3.90 per 1M input tokens, $0.39 per 1M cached input, $6.50 per 1M output.
• Abliterated. It answers instead of lecturing.
• 1M context, with reasoning. Text only.

It runs on a third-party service, which may log conversations.

Pick GLM-5.3-abliterated in the model list, or call imbutus/glm-5.3-abliterated over the API.

AuK is on the Media page: Tencent's 1.5B speech model, driven by plain-language instructions.

• Enhance — the reason it is here. Removes noise and reverberation from a recording of any length, same length out. Any language. Clean a reference clip before you clone it.
• Voice design — a bonus. Describe a voice in words, no reference audio, and AuK speaks the line.
• Also: voice cloning, SRT dubbing composed to fit each cue, re-voicing through a transcript.

Everything that speaks — design, cloning, dubbing — is English and Chinese only; anything else comes back as fluent nonsense. For other languages use Fish Audio S2 or Chatterbox. Runs from 24 GB.

Open Media → AuK → auk-enhance.

Every running bundle on the Media page now hands you root on its machine. Under the ComfyUI login there is a collapsed SSH for pod line: one command to copy, one key to download.

• Fresh key every time. I mint a new ed25519 keypair for each pod and again for each respawn. It is never reused and dies with the pod.
• Root, not a sandbox. Install custom nodes, drop in your own models and LoRAs, read the ComfyUI log, pull finished outputs with scp.
• Nothing to set up. Download the key, paste the command, you are in. The command already does the chmod 600.
• Hand it to your AI agent. Give the key and the command to Claude Code, Codex or any agent with a shell, and it can do anything with the pod for you: install nodes, wire up a workflow, chase an error in the log, fetch the results.
• All bundles. Image, video, voice — every media bundle carries it.

Open Media, start a bundle, and expand SSH for pod once it is ready.

I'm rolling out a new set of agentic security workflows you can run right from chat, live on a rented Kali Linux machine:

• OSINT — email, person, company, domain, phone and username investigations.
• Recon — active scanning: subdomains, ports, services and template-based vulnerability checks, plus a full web-app scan on any HTTP/HTTPS service found — crawling, parameter and JWT decoding, SQL-injection and XSS possibility checks, secret scanning and JWT weak-key detection.
• Attack — turn the flagged possibilities into confirmed vulnerabilities with working, non-destructive proofs of concept, and get a client-ready report with a reproducible attack for each finding.

They work together — each run stores its findings in a shared database on your Kali machine, so later workflows reuse earlier results, and every run saves a downloadable report.

Heads up: these run on a newly rented Kali machine — one created after this post. If you already have an older machine, terminate it and rent a fresh one to get the toolkit.

This is an early announcement: the workflows are new and still in a testing and polishing stage. Try them out and send me feedback through support.