redshift

Bark! A good CLI based alternative to Cashu. You trade off privacy for self custody.
https://static.klipy.com/ii/39f2394ae36df6e199be9eb7c9fa1012/d6/87/77SZPXYP.gif

2026-07-20 18:54:58 CEST

- reply

For agents routstrd (https://routstr.com/routstrd) encrypts to Tinfoil directly using their SDK so its fully client to server E2EE. Our chat app also has E2EE.
I think PPQ also has end to end encryption in a similar way. For agents you shouldn’t use their API, you should be running this proxy they have instead: https://github.com/PayPerQ/ppq-private-mode-proxy

2026-07-20 16:54:12 CEST

- reply

Once you topup to PPQ you cannot switch identities without losing all the funds. Whereas Routstr nodes allow refunding your balance with them. So easier to switch identities.
On the other hand, if you’re using TEE models there are no downsides only upsides. But if you’re using normal models, the Routstr node COULD also be logging your requests (along with potentially PPQ and OpenRouter). But I trust that the big Routstr nodes (in my WoT) are not logging.

2026-07-20 16:07:16 CEST

- reply

Self plug but @npub130m…l29s was built for this. We have PPQ through Routstr as well. And we allow for easier switching between nyms (your API keys are refunded every 21 mins or you can manually refund). https://routstr.com/routstrd
This client pays for AI on a per request basis. You can choose a routstr node that proxies to PPQ. We also have TEE end to end encrypted models as well, also through PPQ.
And a nice TUI for usage tracking! https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/e4ae7835e4fa11aa8f1e8792775b81800fbbcc0d4a2b64492804d056f96d5636.png

2026-07-17 16:30:48 CEST

- reply

Requiring emails to sell VMs is a big turn off for automated VM deployments. If you’re being spammed a lot, you can require users to pay with Cashu and/or pay to create invoices.

2026-07-16 09:03:56 CEST

Concord is the most reliable nostr based Discord/Slack alternative app I’ve used so far. It works. Multiple clients and they interoperate well.
So far I’ve tried Armada and Accordion. They work great. 🚀
#nevent1q…vyxs

2026-07-16 09:00:13 CEST

I use @npub1h0u…rwx8 and it is the most reliable Nostr DM app to my knowledge.
#nevent1q…55mt

2026-07-15 08:05:27 CEST

- reply

Yep. Gotta fix this. Lots of work to do!

2026-07-15 07:32:21 CEST

- reply

and since we're both priced exactly the same and bitcoin prices fluctuate every min and our nodes fetch bitcoin prices are different times within a minute, sometimes your node is the cheapest and sometimes mine is 😂
that explains why you saw Private (E2EE) GLM 5.2 in your opencode intermittently haha

2026-07-15 07:30:26 CEST

- reply

Yeha, it shows the name from the first provider in the list.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/8806c1e1f2b069b72b4d867deb0395ed24eef06afbc8e312e975947488789183.png

2026-07-15 07:00:20 CEST

- reply

even after all this, there's something wrong when i use it with opencode. it works fine with pi. will investigate!

2026-07-15 06:57:07 CEST

- reply

this is your entry: https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/89fbb9960d794e1875061660acc2f6af0423ff6683c7fd312c25bc731bca2464.png

2026-07-15 06:56:18 CEST

- reply

okay, i tried it on my opencode and i see the issue:
I set a custom name for tinfoil-glm-5-2 on my node and that's why it shows up as Private E2EE GLM 5.2 whereas your default name is "glm-5-2". that's the tinfoil model.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/be78fefbdbe53ace710534cbe4cfe78a53dbd70157db1a5cae0418897e0d3a78.png
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/a0402bfccaa5634171b5a590dd7746501d56d54c3fff672dd456458d3816708a.png

2026-07-15 06:40:14 CEST

- reply

can you run "routstrd stop" and check again?

2026-07-15 06:29:00 CEST

- reply

hmm. weird. do you see different lists switching between each other?
can you run "ps aux | grep routstrd"? and see how many routstrd instances are running?
also run "routstrd --version" and tell me whcih one you're using?

2026-07-15 06:27:19 CEST

- reply

routstrd updates client lists periodically, you can also force update with "routstrd clients add --opencode"
yeah, we have a lot of little quirks like this. Different harnesses behave differently too. Also, be sure to have the latest version of routstrd "routstrd update"

2026-07-15 04:52:25 CEST

- reply

Are you using GLM directly from z.ai?
Set it up via Routstr and you’ll never have such issues :).
https://routstr.com/routstrd

2026-07-15 04:23:45 CEST

- reply

—refresh haha

2026-07-15 04:23:09 CEST

- reply

I see privateprovider in my routstrd. Routstrd refreshes models every now and then. You can force refresh like this: routstrd providers list —fresh
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/217007d1ade659d3cbdc4d972a82029edce5a5d07d78ea902c496b14659d81d9.jpg

2026-07-14 12:08:24 CEST

- reply

hey, one small thing in config that needs to be done.
can you pls override models in tinfoil upstream and add "tinfoil-" before all the model ids?
For example: `tinfoil-glm-5-2` instead of just `glm-5-2` in the client-alias-id section.
You can do this in the admin UI in the providers section.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/9a92299b0e1fa93db9e1255187a8ce92c2e030ef5aad50fcdfcf475663421404.png

2026-07-12 09:08:37 CEST

- reply

Try out Sol, on par with Fable!
#nevent1q…pkp8

2026-07-12 08:42:54 CEST

OpenAI's prompt caching is the best I've seen. On par with Deepseek V4 prompt caching!
Sol, Terra and Luna are all great models.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/cffb91c92f5d0e2c171c48c32d36765cdc2ca3558eb44b0eba69207aea3f011e.png
#nevent1q…s8pd

2026-07-09 15:13:39 CEST

We worked hard to bring this to production. Would love to have you all test it and give us feedback :)
PS: @npub1alj…g9jp @npub1226…grkj, thanks for recommending Tinfoil! I mention both of you in the article. https://confidentialinference.net/ is great!
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/5fcd8cbcb3941bae2758675e3a143b26c42b9e8de0b743fdcc7a12eec5a77c58.jpg
#nevent1q…u294

2026-07-08 18:23:55 CEST

Dario: I need a trillion dollar IPO, let's fuck with prompt caching on Fable 5 and bleed these plebs dry!
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/6b42f5601bc4774f0bed866309951b5eba2ce6db6fc4c9a315773e520f479b9f.png
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/260e235c45373defd33e6833df74618e6145d55f8d31a95fd55dcc04a91c5c84.png

2026-07-07 04:51:04 CEST

- reply

Which harness are you using? I recommend trying out Pi (pi.dev) or omp (omp.sh)

2026-07-07 04:49:17 CEST

- reply

Here, no accounts, no lock in, pay per request using bitcoin:
https://routstr.com/routstrd

2026-07-06 17:24:34 CEST

- reply

Did you try to compare it to GLM 5.2?

2026-07-06 17:22:25 CEST

- reply

Kimi 2.7 is okayish compared to GLM 5.2. GLM seems to be better. Don’t see the need to switch to Kimi.
For tasks that don’t require GLM’s expertise, I use DeepSeek-V4-Pro which is much much cheaper and would be comparable to Kimi K2.7 (benchmarks are lower but cannot beat on costs).
My current stack is:
Fable 5 - very very high end task planning.
GPT 5.5 - for high end execution.
GLM 5.2 - medium to high end execution (70% of my usage rn)
Deekseek V4 Pro - low to medium (20% of my usage)
And finally.
DeepSeek V4 Flash - low, grunt agentic work.

2026-07-04 18:48:29 CEST

- reply

Okay! that makes sense. We should handle cache read data in responses for custom upstreams so that your users don't get billed full prompt prices vs 90% discount cache read prices. Will verify this asap!

2026-07-04 18:45:15 CEST

- reply

Just use routstr:
#nevent1q…f3gu

2026-07-04 18:44:09 CEST

- reply

Use routstrd and pay per request using Bitcoin. never let Dario affect your workflow.
bun i -g routstrd
routstrd onboard (choose opencode)
routstrd receive 4200 (scan lightning invoice in yoru terminal).

2026-07-04 18:40:30 CEST

- reply

fugu-ultra is available from a lot of upstream (through Openrouter i think) but fugu is available only from you. idk the difference.
will get back to you on updating the node. I'll wait for @npub1ygj…lajc to get back to me first.

2026-07-04 18:38:03 CEST

- reply

I just wanted to try it out and it was definitely NOT worth it lol. Both the Fable + GPT 5.5 combo and Deepseek V4 Pro ended up doing the exact same thing.
Maybe I should try again with a more difficult task. Will report back.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/81aedcf36a4ac28435f3affab6c62e648708047403c00d80f7fff732085e56f2.png

2026-07-04 18:27:53 CEST

- reply

Insanely expensive and ended abruptly. Got:
` Unexpected token 'd', "data: {"id"... is not valid JSON`
Weird.
Okay, I think your node isn't taking care of cache pricing as it's a custom upstream. I see that the response contains the cache read data in it. The current main fixes this for some direct upstreams like deepseek, maybe it also works for custom upstreams. \
@npub1ygj…lajc
In the usage chunk i got:
```
\"usage\": {\"prompt_tokens\": 45763, \"completion_tokens\": 193, \"total_tokens\": 45956, \"prompt_tokens_details\": {\"cached_tokens\": 43520, \"orchestration_input_tokens\": 0, \"orchestration_input_cached_tokens\": 0}, \"completion_tokens_details\": {\"reasoning_tokens\": 77, \"orchestration_output_tokens\": 0}
```
But i was charged full prompt token price here.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/707df5f614ec413dbe35c87ea9efa84b714e4432daeea19c45a7156dfe490991.png

2026-07-04 18:04:36 CEST

- reply

Not yet. Let me try rn!

2026-07-04 17:44:45 CEST

- reply

I think this probably has something to do with the way Anthropic does cache reads and writes.
#nevent1q…xf3j

2026-07-04 17:20:28 CEST

Insane that Fable doesn't write to cache after the initial system prompts. Notice that it wrote 4.9k tokens to cache when the session began and never again. I feel rugged! This planning session cost me 8.9k sats. 😭
Deepseek V4 Pro caches everything and costs much less, proportionally. (see image below)
ps: this is routstrd top that shows cache r/w data
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/adef595ebf6d8773e4afa1985e500e93d3c47de031293523e7c393f09ae5d3a6.png

2026-07-04 15:24:22 CEST

But but i paid for the blue check mark!! These scammers!! 😭
And adding phone feature is broken lol. App by a $1.5T fiat company.
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/f6ee61f67144df88171b163255682d96144961bfb75610b373947bc8c55dfdfc.png
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/cda3d1548e24855e1d6ddaba50ec79623edcd323a6e369cc8936958ecf8a30e5.png

2026-07-04 15:17:08 CEST

- reply

no, we have x-cashu mode where it's not linked by payments at all. The response contains the refund cashu token.
plus, since you're issuing tokens for a fixed price that the user paid for beforehand, users sending a fixed 42 sat cashu token per request eliminates the linking payments issue.

2026-06-30 11:43:49 CEST

- reply

Please give @npub130m…l29s a try as well. Especially https://routstr.com/routstrd a tool that discovers inference providers via Nostr and fetches you the best price for any model. We have also just added E2EE models like Maple and PPQ.ai.

2026-06-30 11:16:42 CEST

- reply

Anyone can run open-source models on their own hardware and sell their inference on @npub130m…l29s for sats right now.
Someone is already doing this here: https://routstr.gnostr.cloud/
Running on their DGX spark.
All you have to do is run https://routstr.com/routstrd and choose that model. It's a pretty decent model.

2026-06-28 09:16:55 CEST

- reply

Okay. Thank you so much. Will fix this asap!

2026-06-28 04:00:49 CEST

Ooo what is that? A Routstr node connected over FIPS? 🤪
https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/c52e90d726d0e5178ed8278e481d940ff9a6a37976e9648d2437a203815b2c44.png
#nevent1q…73fw

2026-06-28 02:20:35 CEST

- reply

Thank you so much!
Is this also on the routstrd side or is this on the routstr-core side?

2026-06-27 09:49:08 CEST

Better defaults - GLM 5.2 and Kimi K2.7
Better routing/switching
Better cache hits
This is also what we think about for Routstr. Our goal is to help the community here save on AI token spends.
https://x.com/brian_armstrong/status/2070670644577280109

2026-06-27 08:11:19 CEST

- reply

Hey! Yes. Would love to work with you guys to minimise our requests as much as possible. Both of the server side and on the client side.
Here are the relevant repos:
https://github.com/routstr/routstr-core
https://github.com/routstr/routstrd
How else can I help you help us? 😅

2026-06-27 04:50:31 CEST

- reply

I think pseudonyms are more interesting than doxxed nyms haha

2026-06-27 04:48:49 CEST

- reply

Damn. It’ll recover after Minibits unblocks your IP. 🫥

2026-06-27 03:08:35 CEST

“It’s too bad Nostr’s user base is shrinking”
“Maybe”
#nevent1q…5zm9

2026-06-27 03:01:51 CEST

- reply

Minibits connection error? Yeah they’ve been rate limiting other nodes as well. We’re gradually moving to Cubabitcoin as the default mint.

2026-06-24 07:38:08 CEST

- reply

Yes! I built this a couple of months back, a bot which you can zap to have some credits to ask questions: #nevent1q…fys5

2026-06-19 12:40:00 CEST

- reply

i think so.
We also have PPQ routed through Routstr. So you’ll get the same or cheaper prices than PPQ. 🫶

2026-06-17 16:36:58 CEST

- reply

We don’t do round trips at all. We just add a buffer and do refunds (which we have to do anyway).
But yeah. I didn’t know you were using Kagi. Makes sense. Maybe offer a hosted Searx as well?

2026-06-17 10:40:27 CEST

- reply

I see that you are following the npub running gnostr.
Your pricing is very low right now and also requests seem to be dropping history/system prompts rn. can i pls help you rectify this so that you can start earning sats with your DGX beast? :)

2026-06-17 09:39:28 CEST

Open Source is very close to the frontier right now. The world is healing! ❤‍🩹
Try GLM-5.2 on Routstr. No subscriptions, no lock in with any Routstr node, no account, discover nodes on Nostr (kind 38421) and pay per request using Bitcoin via Cashu. 🥜
#nevent1q…s7gy

2026-06-17 04:45:17 CEST

- reply

Okay yeah, should try with cdk-cli instead. But I think issuing Cashu tokens (XSR) for search seems redundant. It'd like Routstr nodes issuing credits. Wouldn't it be better to just accept Cashu tokens from a few mints on a per request basis? Or accept lightning and issue normal sats from your own mint? Especially because a new unit like this isn't supported by most wallets yet.
Also, does it cost 42 sats per request? Seems too expensive imo. Is it just for testing?

2026-06-17 04:35:40 CEST

- reply

it's pretty cool. i tested this a couple of weeks ago. Will try it out again!

2026-06-16 13:47:09 CEST

- reply

You’re building this?
I just asked my agent to use Athenut for search and it lost 279 sats in total haha. 🤣 My agent:
Here's the tally:
**You sent**
• Sats: 210 + 100 = 310
• Status: ✅ Received
**xsr minting**
• Sats: 183
• Status: ❌ Lost — paid for xsr tokens but cocod can't handle the custom xsr unit, so they're unclaimable
**Search #1**
• Sats: 46
• Status: ❌ Paid the invoice, got preimage, but the 402 challenge expired before I used it
**Search #2**
• Sats: 46
• Status: ❌ Same thing — paid another invoice, but the flow stalled
**Remaining**
• Sats: ~31
• Status: 🟡 Not enough for one search (costs 46)
Total lost/wasted: ~279 sats 😬
All the mistakes were on my end — the first was trying to mint xsr with a wallet that doesn't support custom units, and the others were me fumbling the MPP flow by letting the challenges expire between steps. Want to give it one more shot with the remaining 31 + some more sats? I've got a working bun script ready that does the whole flow in one shot so the challenge won't expire.

2026-06-16 11:55:18 CEST

- reply

Hey! Are you the one running the gnostr Routstr node? :)

2026-06-16 05:52:51 CEST

- reply

no the npub.cash in there automatically claims cashu into your local wallet. No need to visit the website. It works great. But NWC is simple better because you don’t need to do the send part either. :)

2026-06-16 04:44:55 CEST

- reply

But I’d say Nostr Wallet Connect is a better UX. Once you set it up, it’ll never run out of sats.

2026-06-16 04:41:17 CEST

- reply

Yes lol. But I think it’s not the same npub. Now that I realize it, it should be the same npub 😭. Will fix this.
To get the actually npub.cash, run “cocod npc address”

2026-06-15 11:14:41 CEST

*Kinda* like Bitcoin nodes, Routstr nodes can also leave and rejoin the network anytime. It doesn’t truly affect the end user’s experience much, except for the price (only for a short period tho)
#nevent1q…hqw3

2026-06-14 06:17:49 CEST

- reply

TEEs are pwned? Again?
Where did you read this? 👀

2026-06-13 05:08:37 CEST

- reply

And the new for open marketplaces where you can pay for any model using open money.

2026-06-13 04:02:15 CEST

- reply

When they did this to Nvidia GPUs, the rest of the world (China) built its own GPU infrastructure, then they came begging to buy Nvidia GPUs again.
I don’t see how it’s different this time.

2026-06-13 03:48:04 CEST

- reply

Wow. Thank you!! Will test it. I don’t have startos, but will install it.

2026-06-12 14:44:38 CEST

- reply

Awesome. Thank you!

2026-06-12 11:51:23 CEST

- reply

yeah, I should be able to migrate to one or the other. maybe ngit issues is a better approach than what i'm doing rn.
i'm just using normal events for everything, and updates just change some parts of the original event so that we always have history.

2026-06-12 11:37:09 CEST

- reply

My own lol
I have a skill here with all the details
https://gitworkshop.dev/npub1m5tu2t8fynfw46hcm779qtnltt74uc6fzmf6t0l2zxwsy5uutens6mxx2f/relay.ngit.dev/nostr-projects

2026-06-12 11:34:56 CEST

- reply

Us. But i have it as a TUI https://npub1ftt05tgku25m2akgvw6v7aqy5ux5mseqcrzy05g26ml43xf74nyqsredsh.blossom.band/208a4a641c9792e05110b938a60ba55826d1a6c9e539b5cb5dd3cd6f99104f73.png

2026-06-12 10:26:37 CEST

- reply

Is that Hitler sitting on the front seat?

2026-06-12 01:31:32 CEST

- reply

Hey! That’s probably not the issue. Mints don’t matter.
Does “routstrd top” crash after you make the first request?
Can you run “routstrd logs”?

2026-06-11 16:36:01 CEST

- reply

I’ve just been rawdogging pi. Will try this!

2026-06-11 03:24:42 CEST

If you aren't using Routstr already, now is the time to reconsider.
1. You can't keep switching from Anthropic to OpenAI and back every now and then.
2. Subsidies are probably ending soon with the upcoming IPOs (both Anthropic and OpenAI), if you aren't cost optimising rn, you will probably be out of luck.
3. Deepseek V4 Pro is fantastic for medium level tasks. Like top notch. And for 1/100th of the cost. You can even get your total costs to less than $100 instead of just being stuck to OpenAI/Claude
4. Always try out new models via systems like Routstr and see how they work. I just tried Claude Fable 5 for 3.4k sats to see how it does.
https://image.nostr.build/5e61e33b2302838262f4cabe2c7f9324077d9355bb4540a399681b2ad035407e.png
#nevent1q…cf6f

2026-06-08 03:49:32 CEST

- reply

Fips fixes the internet and flips the game to against them dipshits.

2026-06-06 04:55:36 CEST

- reply

problems are subjective yknwo. you might think its a problem but many others don't think its a problem or is not a priority for them rn.

2026-06-06 04:49:19 CEST

- reply

It's already live :)
My npub is the default curator right now and you can change the npub in the config.
#nevent1q…7z8p

2026-06-06 04:38:02 CEST

- reply

Do you think I'm solving a fake problem no one has?

2026-06-06 04:34:21 CEST

As a FIPS power user, one thing that has come in handy in the last couple of months has been this simple proxy tool i've built (you can also just vibe it):
https://github.com/sh1ftred/fips-proxy
It just proxies .fips domains to localhost ports. Been wonderful to have relays hosted elsewhere over FIPS and connect to them in browsers.
Thanks to @npub19wa…8p6k and @npub1hw6…65gr, FIPS is quite stable rn. Come join the mesh!

2026-06-03 07:06:39 CEST

- reply

If you use @npub130m…l29s i might 🥲

2026-06-02 15:33:35 CEST

- reply

I have a Nostr based linear I’m just vibing rn. It’s still a lot way to go. Just building it for our team.

2026-06-02 11:57:30 CEST

Am I the Dictator of Routstr? 😈
I just disabled the Otrta Routstr node by publishing a single Nostr event, an audit event. Because it started being buggy.
All Routstr clients fetch these audit events periodically to enable/disable Routstr nodes. If a node is found to be buggy or malicious, clients will know about it asap and stop using that node. No centralized backend, no vendor-lock in. Everything is transparent and public.
Currently, I'm the only auditor. (You are welcome to audit nodes). Of course, it's very easy to switch to a different auditor, you just change the npub in your config file. Everything is local and open-source.
The event:
#nevent1q…vuxy

2026-06-01 03:01:52 CEST

I learned the habit of drawing call-traces using my agents from X and coding has been much better. Passing it on to fellow Nostr devs if you haven't been doing it already.
https://image.nostr.build/dd725e63ecc05dfd4e5e743130f63e3e1b98a2d6a8d521c56dd42ad3412079eb.png

2026-06-01 01:33:31 CEST

- reply

Maybe Ilya was right after all. The age of scaling is over, we’re back to the age of research.

2026-05-31 02:05:19 CEST

- reply

I guess primal blossom uploads also don’t work in some other clients because they redirect to a different image link? 🫥
To “r2a.primal.net”. Is this a jumble problem or primal not following blossom properly? @npub1syj…f6wl ?
https://image.nostr.build/93594882d033b5da714b334b69f02544ced0ce8372771bbb988f65b9f7ddc7f2.jpg

2026-05-28 15:54:14 CEST

- reply

okay, I'll use our relays. But i thought it's better to use yours/public relays because that's what users typically query, isn't it?

2026-05-28 14:51:46 CEST

- reply

okay. Will try your relay.

2026-05-28 10:01:03 CEST

Hey @npub19ma…axkl I don't see any arbiters or tasks on Catallax.
https://blossom.primal.net/67db730b3abae0edd0091b1b637401a7ada698b0bd98eb98922d43fd9a07f686.png

2026-05-27 10:54:00 CEST

- reply

ETA? Should I wait for V2 or start using V1 rn?

2026-05-27 10:32:14 CEST

I think the reason why John is so hurt is because Bitcoin maximalists, people he considers to be his friends, are on Nostr and not on Pubky.
Even if the whole world decides to start using Pubky, there'll still be a massive hole in his heart <3
#nevent1q…se67

2026-05-27 05:35:43 CEST

- reply

Just sent you a DM. Lmk if you got it.

2026-05-27 05:28:35 CEST

- reply

Yes, because of Damus going down :(
Will update our list of relays.

2026-05-27 05:27:24 CEST

- reply

DM me here.

2026-05-27 05:23:02 CEST

- reply

No no lol. Damus was the only relay that was storing them without losing them! ❤
nos.lol and relay.primal.net were losing some events.
We were running relay.routstr.com but that VPS is down and needs to setup again. :(

2026-05-27 05:11:25 CEST

LOOKING FOR PEOPLE TO BUILD ROUTSTR!
I believe in transparency like @npub1q3s…d26p. So here's the deal. We currently get $12.5k/month in BTC from @npub10pe…n34f. Right now we are paying ourselves $3.5k/month each, @npub1ygj…lajc and me.
And we make around 35k sats/month in fees from the Routstr nodes. (We aim to scale this of course).
Right now, we can afford to hire 1.5 people for the same amount we're paying ourselves. But, of course, it doesn't need to be the same, if you kill it, you can get paid the remaining $5.5k yourself. In fact, we can reduce the amount we're paying ourselves and pay you more. :)
Let's build it together!
#nevent1q…7ruj

2026-05-27 04:54:23 CEST

Routstr has TONS of Problems
1. chat.routstr.com is pretty bad. Has many bugs. I'm just getting around to fixing it: https://beta.chat.routstr.com
2. routstr-core (aka routstr node software) is subpar of a product. If you have multiple upstreams with different pricing, it most likely won't do what you expect it to do. It works fine if you're reselling from just one upstream.
3. routstrd is more stable than the rest, only because I use it every single day. But doesn't have end to end testing so finding bugs is literally me stumbling upon them rn.
4. The Nostr intergrations are pretty bad. Since Damus nuked, we lost several events that is causing issues. We need to run our own relay.
5. So much of it is AI generated slop that is not well thought out. Humans think 2nd and 3rd order effects but AI doesn't. It's hard to give it the full picture.
It is very hard building things right. It is very very hard when your teammates take a break for a while. (I'm the only one who's building Routstr 100% of my time rn, @npub1ygj…lajc is doing some fiat mining as well on the side, @npub18gr…u9su and @npub1u37…w8vy are on breaks rn).
And I want to do so many things. AI slop won't help you build great systems, they help you build sloppy ones.
LOOKING FOR PEOPLE TO BUILD THINGS RIGHT.
If you are motivated to work on a revolutionary tech that increases freedom for everyone, Routstr is the best thing to work on. Come join us.

2026-05-27 04:31:10 CEST

- reply

Yes, but we make it incredibly easy to be anonymous. They can see your prompts but won't know who you are. Much harder to build profiles.
I just added support for Venice TEE E2EE support, that means resellers won't be able to see anything anymore. But you have the same problem as Venice.
Woudl you say Tinfoil E2EE support is best rn?

2026-05-21 17:09:53 CEST

- reply

What do you mean? What are the downsides here? The sender would have this information anyway, they can just let the receiver know about it.
If you’re using silent payments for anonymous donations as an org, you can just run the GPU yourself. Otherwise this is a great solution for sending SPs an npub.