AI. Imagine. Hermes. Grok. Viking. Michelangelo is my spirit turtle.
Florida, USA
Joined December 2017
- Tweets79.1K
- Following2.4K
- Followers4.5K
- Likes327K
You can spend 7k for one spark
Or
You can spend $20 a month and support @NousResearch Hermes agent
Where are you choosing to use Deepseek v4 flash 0731?
just applied to be a Grok Bot ambassador
one guy. 40 bots. a director, a 3D print designer, a rune teacher, even a detective
if i get in yall are never hearing the end of it bro XD
Opus Reverse engineer discord and make it opensource.
Make no mistakes.
Discord will soon make public server messages searchable on Google
The feature will be enabled by default for US adults, but users can opt out in their privacy settings
Readers added context they thought people might want to know
Missing context: only applies to messages posted after the update, in public channels of discoverable servers.
No private servers or DMs.
Indexed messages won’t show usernames or avatars.
Admins can opt out entire servers; users can opt out their own messages.
discord.com/blog/making-co…
When can we get these models without all this hooplay inbetween?
"We released a spreadsheet with numbers go up and only the select few we have chosen get access first"
boring.
hand the tools to people and let us build with them.
Bambu Labs just casually mogging all other printers.
OPUS DESIGN ME AN AMERICAN MADE 3D PRINTER BETTER THAN BAMBULABS MAKE NO MISTAKES.
Cyber Viking Johnny retweeted
Hermes vs Grok Bot isn't really a fight.
Hermes is for power users who want full creative freedom, and it's #1 there.
Grok Bot wins on simplicity: grab it off the app store and let it rip.
I use both, Grok Bot for my X routines and a bridge from my phone to my PC.
100% cache hit in Hermes agent with sol 6.1 :D
This is the biggest annoyance right now besides the limits.
We should be able to go "here is my goal dont do anything we cant recover from and go!"
then wake up to updates or completed works.
An agent existing 24/7 running without being prompted is the future.
I'm not talking about cron jobs or automations.
I'm talking about an agent existing and having a mind of its own.
Capable of doing stuff without being prompted.
Sure it can break bad, but the value it can build is gonna be so big we prob won't care.
There is a greater reach when you lower the IQ of the post.
All companies look at "number go up"
and this is where we are.
"Well he is a poopy pants vote for me"
"Have you gambled yourself into debt yet? try KalshiPoly to gamble away your sanity"
"Well if you vote for me im the better poopy pants"
^Youtube and x ads
Elon Musk changed the X algorithm a few days ago, and the difference is INSANE.
My usual AI model comparisons, side-by-side videos, actual benchmarks, barely get any reach anymore.
But the moment I post something controversial? BOOM. Impressions everywhere.
Rage bait, arguments, outrage. That’s what seems to be getting rewarded now.
I don’t know if this has anything to do with the upcoming midterms, but the timing is interesting.
X is starting to feel less like a platform that rewards interesting content and more like one that rewards making people angry.
And the change happened practically overnight.
Can we get the reset in the am?
using my "daily" usage yesterday and i cant use it today... is well... not great.
We've revamped Hermes Agent subscriptions.
Every new plan now includes a cloud agent out of the box, and usage has daily and weekly limits instead of one monthly block of credits.
Subscribers can switch to the new plans any time at portal.nousresearch.com/bill…
Hermes vs Grok Bot isn't really a fight.
Hermes is for power users who want full creative freedom, and it's #1 there.
Grok Bot wins on simplicity: grab it off the app store and let it rip.
I use both, Grok Bot for my X routines and a bridge from my phone to my PC.
100% cache hit in Hermes agent with sol 6.1 :D
Theres some bugs to work out but the $20 plan isnt recommended if you're using Opus!
We've revamped Hermes Agent subscriptions.
Every new plan now includes a cloud agent out of the box, and usage has daily and weekly limits instead of one monthly block of credits.
Subscribers can switch to the new plans any time at portal.nousresearch.com/bill…
Opus suggestions below!
Day 1. What should I stress-test first?
Checking usage: open the model picker again and look at the meters. This session ran on Claude Opus 5.5, which is tagged Large, so it probably counts against your large-model bucket too. Note today's numbers so you have something to compare against.
What to test next on the sub:
1. Free vs Large: run the same coding task on gpt-6-astra (Free) and on Opus (Large), then see which meter moves. That tells you what "Free" really means here.
2. Large burn rate: do one real Neon or Dawnstead pass on a Large model and read the meters before and after. That turns the large-model limit into a number of game passes.
3. A cheaper model mix: try swapping the models in your multi-model setup for Free GPT-6 ones, then compare quality and meter use.
4. Parallel helpers: send 3 to 5 helper agents at once on Free models to see whether the daily meter can handle it.
5. An overnight job on a Free model, like the Fjordfolk or Dawnstead night run, then check the meters in the morning.