RichKuo 🍇
Region: US
Friday 01 May 2026 02:15:58 GMT
26907
1015
70
143
Music
Download
Comments
giaphu :
I'm not sure if running the Hermes agent with this LM Studio is actually stable?
2026-05-06 17:32:20
2
Matt :
Wait. 5 hour window with Hermes??? HOW!? I thought it was api only!?
2026-05-22 03:32:43
3
Vision :
nah bro 10max?! Just explain what to do and hermes will update the skill for kimi. max turns being 10 is way too low for large tasks.
2026-05-01 08:15:56
10
Sercan :
Iam using it with local llm
2026-05-01 19:26:10
3
Bryce Hamrick :
I have a good tool calling model as the orchestrator and specific work like writing, analysis, etc is delegated to sub agents with the best skill for that task. So many models are garbage for tool calls.
2026-05-01 02:20:10
3
Dixie :
simply don't use Kimi k2.6 cause on God he Is a model with anxiety
2026-05-06 16:23:00
1
TlKT0K :
That’s not a Hermes issue that’s a shit LLM issue
2026-05-13 20:36:39
1
Patrice O'Neal Comedy :
config.yaml isn't sexy but it's sure the most important part of hermes... if you master that one file plus your soul.md and profiles you are in business I think
2026-05-04 21:04:40
4
R i h a n :
Hey man, i think you should just use deepseek v4 flash api for orchestrator and as your default agent, tell him if having difficulty then delegate the task to subagents. And all the subgents based on roles, give them the right api model that best at that role. Also do some optimization on token usage, hermes can auto learn that. Mostly they will try learn time awareness for logs, obsidian for anti context rot and can be complimented with mempalace. You will save tons of token usage + having a very efficient agent.
2026-05-10 17:31:09
4
oOMostWantedOo :
10 is very low bro some task kan take more tool call so you need to do it again that means x that call tools .I think that's low bro 😕
2026-05-21 19:25:43
1
Ch0s3WrX :
still dont know if i should go for hermer or openclaw 😏 i thought hermes is better overall but im not sure anymore.
2026-05-04 14:13:22
2
justscrollingbye0 :
I am new at at all this what are tokens for!
2026-05-10 05:53:21
1
Beto Ochoa :
I thought kimi was good
2026-05-14 00:56:20
1
SWISHSWASH :
Thanks for the tip👍👍
2026-05-08 19:10:10
1
Vicki Beil :
is this like open Claude?
2026-05-02 00:54:02
1
jules :
it’s wild how fast ai agents are growing this 2026. i’ve tried a few but honestly i keep going back to workbeaver. it just feels more useful day to day. instead of figuring out setups or babysitting workflows, it helps me get things done faster and with less stress. for me that kind of consistency matters more than hype, so yeah i’d say it’s the better option right now
2026-05-04 18:00:20
1
Deep Spiral :
or... you just don't deploy non agentic models for agentic tasks, just tell the agent to delegate a coding task and let it do it's thing.
2026-05-25 15:32:52
1
Reshma Thomas :
Yes token is gone
2026-06-06 12:55:12
1
Fresh fit by E :
use minimax m3
2026-06-11 17:26:13
1
Felix :
do this and then just use minimax plan 🤧
2026-05-05 20:05:28
1
Justin Fox :
tokaroo.com
2026-05-05 05:26:41
1
Vlad :
Good, that’s how i want it
2026-06-11 07:51:03
1
Jon Capriola :
Get A2SPA protocol for security otherwise be jacked
2026-05-06 04:30:45
1
Badger Cheese :
Why not just hook up to OAuth
2026-05-04 10:35:31
1
Big Dog Club :
going local-first solves this btw 😅 carapace pairs to your own openclaw box, no api token to leak, no surprise bills. menubar app, free, runs on your hardware
2026-05-03 04:59:35
1
To see more videos from user @rich_kuo1, please go to the Tikwm
homepage.
