ngmi

https://metaend.qstorage.quilibrium.com

hi~

Public Key

npub14p7fhlz9wrlptt7a5mqqjkk79460k98r999l7dafv4fznqaa565qhjdzry

NIP-05 Address

Profile Code

nprofile1qqs2slyml3zhpls44lw6dsqftt0z6a8mzn3jjjllx75k253fsw76d2qpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqjvamnwvaz7tmwdaehgu3wwd6hqetjveexjetwv3ejummwd35kuef0d4shycnvv50ce8v8

Publishing to

relay.damus.io

nostr.superfriends.online/marble

relay.sovrgn.co.za

relay.nostr.net

purplepag.es

Author Public Key

npub14p7fhlz9wrlptt7a5mqqjkk79460k98r999l7dafv4fznqaa565qhjdzry

Show more details

Last Notes

2026-08-30 10:41:10 CEST

Chat can now create polished, downloadable PDF documents with headings, paragraphs, lists, tables, and page breaks. PDF generation runs locally in your browser, has no separate fee, and does not use a third-party conversion service. Approve the tool when chat proposes a document, then download the finished PDF directly from the conversation.
https://nano-gpt.com/subscription/kUGJhgF9

2026-08-30 10:19:14 CEST

Can you send 1btc through LN or is this impossible?
#asknostr

2026-08-30 09:28:05 CEST

Wen nothing phone support @nprofile…xdyf?

2026-08-30 09:20:57 CEST

Anthropic is cutting Claude Code's current weekly limits by 17%
https://www.bleepingcomputer.com/news/artificial-intelligence/anthropic-is-cutting-claude-codes-current-weekly-limits-by-17-percent/
#claude

2026-08-29 19:15:04 CEST

- reply

Cool

2026-08-29 16:39:12 CEST

- reply

Sounds awesome 👏

2026-08-29 14:33:04 CEST

- reply

Relevant to my interests. Asked myself the same

2026-08-29 10:49:03 CEST

GLM 5.3 and GLM 5.3 Thinking are now included in the NanoGPT subscription at 2x included-input usage. GLM 5.3 is built for long-horizon coding, autonomous agent workflows, tool use, and complex engineering tasks.
https://nano-gpt.com/subscription/kUGJhgF9
#glm

2026-08-29 10:39:30 CEST

What's the decentralized hugging face?

2026-08-28 20:47:16 CEST

Never give your government any info about your holdings. It's too dangerous
#nevent1q…6j0l

2026-08-28 17:20:07 CEST

Why does Marc Zuckercunt want me to disable VPN for syncing images via Bluetooth? Fuck off mate

2026-08-28 15:06:07 CEST

- reply

Super cool tho

2026-08-28 13:52:04 CEST

GLM 5.3 Flash TEE is now available! This private TEE version combines GLM 5.3 Flash's efficient long-context reasoning, coding, tool use, structured output, and image understanding with confidential inference inside a Trusted Execution Environment and attestation support.
https://nano-gpt.com/subscription/kUGJhgF9
#glm

2026-08-27 14:20:52 CEST

- reply

Still a good investment, 2 will probably always let you run frontier open llm 🤔

2026-08-27 14:19:47 CEST

- reply

Dgx spark is around 5k, that should do I guess?

2026-08-27 12:48:23 CEST

Doubt
#nevent1q…48pt

2026-08-27 11:34:17 CEST

Cooking a fully offline PWA note taking app with NIP17 + post quantum security nostr syncing. Can also be used to send pq secure DMs to enabled accounts. Soon
#nostr
https://blossom.primal.net/a0919e82afc1a6bfd58b313b204460192df73f5a12f0452ca1b34c9c667f118a.png

2026-08-27 10:56:50 CEST

- reply

Thanks! Added to our todo 👋

2026-08-27 09:02:17 CEST

Alibaba’s Qwen Team Releases Qwen3.8-Flash-Next: A 125B Multimodal MoE With 6B Active Parameters Previewing the Qwen4 Architecture
https://www.marktechpost.com/2026/08/26/alibabas-qwen-team-releases-qwen3-8-flash-next-a-125b-multimodal-moe-with-6b-active-parameters-previewing-the-qwen4-architecture/
#qwen

2026-08-27 08:34:10 CEST

- reply

Is there a git by chance?

2026-08-27 07:00:51 CEST

AI secops is wild now. Our system found many vulnerabilities in an open source nostr project using Claude and ox-alpha. Devs will have a lot of fixes to prepare in the future
#nostr

2026-08-26 18:34:23 CEST

- reply

Is this the trust me bro eval?

2026-08-26 14:49:02 CEST

Qwen 3.8 27B Uncensored and Qwen 3.8 27B Uncensored Thinking are available again alongside the Obliterated variants. Choose the original Uncensored finetune for stronger instruction following or the Obliterated finetune for a different low-refusal style. All four variants remain included in the NanoGPT subscription at 1x included-input usage.
https://nano-gpt.com/subscription/kUGJhgF9
#qwen

2026-08-26 08:05:15 CEST

- reply

They are

2026-08-26 08:01:48 CEST

Cooking https://metaend.blossom.band/dd498f7c8bb17aa2fb0388e91872cbd7efb86d7c89d290b33277c62d53c3d386.jpg
#nostr

2026-08-24 14:37:29 CEST

The NanoGPT text API is now available as a Tor Onion service. Requests travel through Tor directly to NanoGPT without leaving the Tor network through a public exit relay. Use your regular NanoGPT API key with the supported `/api/v1` routes.
https://nano-gpt.com/subscription/kUGJhgF9
#tor

2026-08-24 12:45:54 CEST

- reply

Stake Venice tokens and you have it basically

2026-08-24 11:03:34 CEST

Why is there no notes function in #amethyst 🙏

2026-08-24 09:21:16 CEST

- reply

I see. So arweave is better for this case

2026-08-24 09:03:37 CEST

Are #nsites as unrugable as arweave sites?
#asknostr

2026-08-23 23:21:30 CEST

- reply

Pretty smart to usd arweave and XMR like that. Wow

2026-08-23 18:26:59 CEST

Big
#nevent1q…udrp

2026-08-23 17:41:28 CEST

Burned through 2 #claude plans this week for work, our infra project. Testing GLM5.3 now with a cheap sub, like it! Haven't done coding, but for cowork stuff it's more than capable.
https://z.ai/subscribe?ic=D9Z5HGTW98
#glm

2026-08-23 17:37:41 CEST

@[email protected] oi

2026-08-22 16:13:00 CEST

- reply

Yep sad reality. When Trump took over I moved all my normie ETFs and stocks to EU € only.

2026-08-22 15:52:14 CEST

Qwen 3.8 27B Uncensored is now available and included in the NanoGPT subscription at 1x included-input usage! This open-weight FP8 model is tuned for fewer refusals across chat, coding, tool use, image understanding, and long-context work. Its subscription inclusion may be temporary and can change with availability and capacity
https://nano-gpt.com/subscription/kUGJhgF9
#qwen

2026-08-21 20:02:33 CEST

Meh figured
"Devuan stays the metal. If Omarchy is tempting, put it on the ThinkPad or in a VM as a daily-driver experiment and keep it away from anything that touches keys."
I really want to belief in arch, but for my needs it's too basic

2026-08-21 17:33:51 CEST

- reply

Also on @nprofile…gs7e

2026-08-21 17:31:57 CEST

- reply

Yes, for selecting an even dumber name

2026-08-21 17:30:23 CEST

- reply

Thanks

2026-08-21 15:18:48 CEST

How do you all buy LN with monero?
#asknostr #monero

2026-08-21 11:49:23 CEST

We ran an agentic security scanner over 96,000 lines of someone else's Rust
What Visa's open-source SAST harness found, what it got wrong, and why this post contains no bug list
https://paragraph.com/@metaend/we-ran-an-agentic-security-scanner-over-96000-lines-of-someone-elses-rust
#visa #rust #aisec

2026-08-21 10:59:16 CEST

👏stack👏sats👏
#nevent1q…wqcm

2026-08-20 14:21:16 CEST

- reply

Armada

2026-08-20 08:52:18 CEST

- reply

No linux? Lul

2026-08-19 20:59:43 CEST

27s added for a total reduction of 55% RAM is insane. Tasked my harness to reduce the load on a fullstack project because I needed it to work on my laptop for travel.

2026-08-19 17:44:57 CEST

🤡🤡🤡
#nevent1q…27h9

2026-08-19 11:46:42 CEST

- reply

btw how do i trash a mail? am I blind or is there no button for it :)

2026-08-19 11:45:54 CEST

- reply

arrived via desktop, seems like its amethyst then. thanks!

2026-08-19 11:36:58 CEST

- reply

External mail to my mailstr, using amethyst browser (but will check from desktop in a second)

2026-08-19 09:35:50 CEST

@nprofile…n7g2 I sent a mail to my mailstr as test, like 1h ago, but my inbox remains empty. 🤔

2026-08-18 19:54:33 CEST

- reply

AFAIK OTS doesn't really work on ethereum because of some limitations (don't remember which, but I was looking into Arb/Base too instead of Bitcoin, which takes up to 24h)

2026-08-18 16:03:54 CEST

- reply

Super cool. I recently built a OTS for a big infrastructure pilot, and it was painful (even with opus5). Reference modules outdated or buggy, so had to do it all. Thinking about releasing a lib for it

2026-08-18 15:04:21 CEST

Running npub1u280zakdwcv23nqpgatsmz8lw6aj22aa6gjjlx0d50lqr8m8xucq7kf93q as a secure backup to this nostr account. in case something happens to it, this is the one on https://www.inkan.cc/users/npub1u280zakdwcv23nqpgatsmz8lw6aj22aa6gjjlx0d50lqr8m8xucq7kf93q - seed is offline, signer is rotateable.

2026-08-18 12:38:00 CEST

- reply

yep thanks. i got used to it

2026-08-18 12:15:17 CEST

- reply

sent you the file for sponosr :)

2026-08-18 12:09:28 CEST

- reply

very cool. I thought about evm for seed contract too. you actually built it. testing now

2026-08-18 08:16:38 CEST

Farcaster is for sale, kinda sad to see it go down so bad
#farcaster
https://metaend.blossom.band/4a889181abcf444d3cddf8505139408636926fd30230c5b60d99e10302728473.jpg

2026-08-18 08:03:16 CEST

- reply

This is the way

2026-08-17 17:15:04 CEST

Microsoft says #GitHub is down worldwide, with the GitHub website, API, Actions, Pull Requests, and other services impacted; the outage started at 13:40 UTC (Mayank Parmar/BleepingComputer)
#ngit is up btw

2026-08-17 14:35:16 CEST

DeepSeek V4 Pro 0813 TEE and DeepSeek V4 Pro 0813 Thinking TEE are now available! These private TEE versions bring DeepSeek V4 Pro 0813's coding, tool use, cybersecurity, automation, and long-horizon agent capabilities into a Trusted Execution Environment with attestation support and a 1M-token context window. Use the regular version for faster direct answers or the Thinking variant for harder reasoning-heavy work.
https://nano-gpt.com/subscription/kUGJhgF9
#DeepSeek

2026-08-17 13:54:50 CEST

Should've used TEE models
https://ppq.ai/invite/9bc57632
https://nano-gpt.com/subscription/kUGJhgF9
https://venice.ai/chat?ref=ksiwS7
#nevent1q…q0sx

2026-08-16 16:39:22 CEST

Imagine you could shitpost mine @amethyst coin by triggering the pow mining option on the posts

2026-08-16 11:21:50 CEST

- reply

It is what it is?

2026-08-16 07:43:43 CEST

Incognito chats now support image uploads! Images are kept temporarily in memory instead of being saved to your Library or persistent browser storage, and are cleared when you leave the incognito session. Other file types still require a regular chat.
https://nano-gpt.com/subscription/kUGJhgF9

2026-08-15 15:14:35 CEST

Super cool tool to check what LLM actually would run locally.
try without installing anything:
uvx llmfit
#llm

2026-08-15 09:19:21 CEST

Testing https://mycircl.xyz/

2026-08-15 09:10:58 CEST

That's cool
https://bsky.app/profile/commandcode.ai/post/3mt3pfzerfa2a
#glm
https://metaend.blossom.band/b5da2013fe5e75d9530770f54c4b084dab20e840c59de223057e1924a74b72d4.jpg

2026-08-14 09:55:03 CEST

GLM5.3 👏

2026-08-13 21:19:37 CEST

- reply

I've been testing metas new model locally. Kinda stupid tho, no idea why it's praised so much. Or skill issue on my side

2026-08-13 15:45:15 CEST

Error: Anthropic stream error (overloaded_error): Overloaded 🤬

2026-08-13 14:39:40 CEST

Anything that's not a website or smartcontract should be done in rust. AI is very good at it, and you get a natural layer of protection on top.
#nevent1q…j935

2026-08-13 08:53:00 CEST

Definitely not a scam pinky promise ¯\_(ツ)_/¯
#nevent1q…9f3s

2026-08-12 16:15:19 CEST

Agent ready C2PA remover - https://gitworkshop.dev/[email protected]/c2pa-remover
#c2pa

2026-08-11 19:04:03 CEST

- reply

Sad

2026-08-11 18:21:30 CEST

I built a C2PA remover over a year ago, with updates this will easily remove the watermarks from claude files
https://github.com/ngmisl/C2PAremover
#c2pa

2026-08-11 11:51:14 CEST

can you run a #nostr relay on a smol rpi with solar?
#asknostr

2026-08-11 08:15:22 CEST

- reply

Sounds super cool. We use armada atm for most stuff, obviously not optimal. This could be a great addition. Any ETA?

2026-08-09 08:52:00 CEST

Neuralwatt is now available in provider selection for Kimi K3, Qwen 3.6 35B, and DeepSeek V4 Flash 0731, including thinking variants where applicable. Neuralwatt cached-input usage now costs 10% of its regular input rate on supported models, except DeepSeek V4 Flash. Neuralwatt has retired its Kimi K2.6 and Qwen 3.5 routes, so those models are no longer offered through Neuralwatt provider selection; both models remain available through other providers.
https://nano-gpt.com/subscription/kUGJhgF9
#ai #neuralwatt https://metaend.blossom.band/3f3fde1ccdbc7e63e95179f589542330415a0685f916daa99bfcd2552b326be4.jpg

2026-08-09 08:33:54 CEST

Some research on investing in simplex wefunder (POV ROI)
Can simplex the company every reach $40m?
**Scale gap:** WhatsApp has roughly 3 billion monthly active users. SimpleX Chat has 430,000+. That's about a 7,000x difference.
**WhatsApp's money (Meta doesn't break it out cleanly, but recent figures):**
- WhatsApp's paid business messaging crossed $2 billion in annualized revenue by Q4 2025 [WizMessage](https://wizmessage.com/blog/whatsapp-business-statistics) , with Meta's Family of Apps "other revenue" reaching $801 million in Q4 2025 alone [WizMessage](https://wizmessage.com/blog/whatsapp-business-statistics)
- Separately, Click-to-WhatsApp ads generated an estimated $12 billion in 2025 [Revenue Memo](https://www.revenuememo.com/p/how-does-whatsapp-make-money) , though that's counted as Meta's ad revenue, not WhatsApp's own line
- Analysts project business messaging could grow to over $8 billion annually by 2030 [Revenue Memo](https://www.revenuememo.com/p/how-does-whatsapp-make-money) if trends hold — which would still be a fraction of Meta's overall revenue
**On the "how likely" question** — I can't predict this with any confidence, but here's context to weigh it against, not a recommendation either way:
- SimpleX's own numbers are telling: they estimate 10,000 paid names at ~$50/year would generate $500K — just barely covering their current $360K/year operating cost. That's the *modest-success* case for one revenue line, not a breakout one.
- All three planned revenue streams (public names, business messaging, paid servers) are pre-launch or proof-of-concept stage — none are generating real revenue today.
- Comparable privacy-messaging projects: Signal is still donation-funded after a decade and hasn't solved monetization. Telegram took roughly ten years and hundreds of millions of users before Premium subscriptions became meaningful.
- More broadly, most Reg CF/SAFE community-round investments never reach a liquidity event (acquisition or IPO) that lets the SAFE convert into anything — that's a structural risk of this investment category, not specific to SimpleX.
- On the positive side: the growth is genuinely organic (no paid marketing), the privacy thesis is hard for incumbents to copy for the reasons the pitch lays out, and the round has drawn credible backers like Jack Dorsey and Asymmetric Capital Partners.

2026-08-08 11:19:43 CEST

- reply

🤔🤔🤔

2026-08-07 12:43:35 CEST

- reply

Because they subvention sub's still. Once this ends, Chinese will be much cheaper

2026-08-06 18:46:43 CEST

If you're a revolut client paid client, you can claim 6 months of chatgpt go for free now as perk

2026-08-05 08:23:56 CEST

Qwen3.8 Max is now available! Qwen's 2.4T-parameter flagship combines advanced reasoning with strong coding, knowledge work, full-stack development, data analysis, and long-running agent workflows. It accepts text, image, and video input and supports tool calling, structured output, selectable thinking effort, and a near-million-token context window.
https://nano-gpt.com/subscription/kUGJhgF9
#qwen

2026-08-04 17:37:29 CEST

Runs
#nevent1q…67v9

2026-08-04 17:35:34 CEST

Insane
#nevent1q…82nk

2026-08-04 08:54:58 CEST

I love OTS, but its slow. Researching if I can use our implementation with #Arbitrum #EVM

2026-08-02 07:55:11 CEST

#amethyst please let me filter out concord channels in dm ui, its too much

2026-08-01 20:45:29 CEST

- reply

I hope this will be figured out sooner. 2030 is pretty close, maybe even earlier

2026-08-01 18:18:20 CEST

- reply

I have to check this out. Sounds interesting. Is this post quantum secure too?

2026-08-01 18:14:49 CEST

Kimi K3 via TEE is live! Use Kimi K3 with confidential inference inside a Trusted Execution Environment while retaining its long-context reasoning, coding, tool use, and native image/video understanding.
https://nano-gpt.com/subscription/kUGJhgF9

2026-08-01 08:10:12 CEST

DeepSeek V4 Flash 0731 is now available and included in the NanoGPT subscription! This re-post-trained 284B Mixture-of-Experts model activates 13B parameters and is built for coding, reasoning, tool use, and long-context agent workflows with a 1M-token context window. ⚠️ Requests go directly to #DeepSeek , so privacy and logging guarantees are limited.
https://nano-gpt.com/subscription/kUGJhgF9

2026-07-31 20:49:12 CEST

TFW the doctor is balls deep into your channel and the nurse tells him about a coldcard hack
#nevent1q…frg3

2026-07-31 20:27:01 CEST

- reply

Is there a shiny version?

2026-07-31 19:08:34 CEST

How often got wasabi or sparrow exploited? 🤔 how often did ledger lose customer data?

2026-07-31 12:14:07 CEST

A hardware wallet that never touched the internet just lost 594 BTC.
Per Coinkite's own advisory on 30 July 2026, roughly 594 BTC (about $38M) was swept from around 500 single-signature Coldcard wallets in about 25 minutes.
The cause was not the airgap. Seed generation fell back to MicroPython's software PRNG instead of the device's hardware RNG. On Mk3 firmware from 4.0.1, that left seeds with roughly 40 bits of entropy instead of 128.
Do the arithmetic, because it is the whole story. 40 bits is 1.1 trillion candidate seeds. BIP-39 runs PBKDF2-HMAC-SHA512 at 2048 iterations, so one high-end GPU grinds about 2.4 million candidates a second. That is 5 days on a single card, under 2 hours on 64 of them, and about $60 of rented compute.
And you only pay that once. Precompute the space, derive the addresses, keep the table. After that every affected wallet is drained on sight. The 25 minute sweep was not the attack. It was the receipt.
Airgapping defends against exfiltration. It does nothing whatsoever about a broken random number generator. Offline is not a security property, it is a threat model, and it says nothing about whether your keys were unguessable in the first place.
Worth keeping honest: Mk4 and Mk5 before 5.6.0, and Q before 1.5.0Q, got about 72 bits. That is 4.3 billion times harder than 40 and nobody is cracking it today. But it is not the 128 you were sold. That is margin, quietly spent, and you were never told the balance.
If you hold one: firmware fixes generation, not existing seeds. A weak seed is weak forever. Generate a fresh one and move the funds. Wallets behind a BIP-39 passphrase appear largely insulated.
The property you cannot see is the one that fails. Verify, do not assume.
#Bitcoin #InfoSec #Cryptography #HardwareWallet #SelfCustody

2026-07-31 11:39:08 CEST

- reply

holy shit

2026-07-31 08:07:52 CEST

Solid
#bjj
https://metaend.blossom.band/9ffa598684f75ec71dcf50d16a13f68ee5165e2e1a2c1a0f514edf0191e94c09.png

2026-07-29 15:11:19 CEST

GM @npub17dh…58ut

2026-07-29 11:52:43 CEST

- reply

i use it, thanks for the service!