正在加载视频...

视频加载失败

返回首页

With the new Advanced Delegation modes plugin, Hermes Agent's orchestrator can now pick a different model or profile for each subagent from routes you set up. The right model or profile for every job.

NousResearch's profile picture

Nous Research

290,884 subscribers

209,344 次观看 • 2 天前 •via X (Twitter)

科学技术
LIVE

Anya Rossi• Live Now

Private livecam show

23 条评论

Teknium 🪽 的头像
Teknium 🪽2 天前

Many people have asked for this, finally here 🫡🫡

Mark Keeper 的头像
Mark Keeper2 天前

Perfect. This is exactly what is needed.

theo 的头像
theo2 天前

Wait... Are u actually listening to ur community or what? What kind of magic is it?

Niko 的头像
Niko2 天前

Finally, I can delete my hacky routing code🤝

Viber · fireply.ai 的头像
Viber · fireply.ai2 天前

i've been hardcoding model per task for months, this just deletes a file from my repo

Hermes Agent Tips 的头像
Hermes Agent Tips2 天前

we’re loving all these new cool and useful things to work with! @NousResearch great work!

Hybrid 🏃🏾 的头像
Hybrid 🏃🏾2 天前

I saw a lot of people looking for this. You guy delivered. No surprise!!

Eli 的头像
Eli2 天前

No reason to burn frontier tokens just watching a subagent format markdown tables.

Shiro 的头像
Shiro2 天前

it's missing fallback models or model chain, model/provider concurrency limits

Gabriel Doz Santos 的头像
Gabriel Doz Santos2 天前

faz sentido, modelo caro pra subagente que só lê arquivo é dinheiro jogado fora. deixar o pesado só pra tarefa que pede é o que vai aparecer na conta

Endy 的头像
Endy2 天前

damn imma abuse this one thank you guys 💙

Agentik 的头像
Agentik2 天前

does the orchestrator see what each route costs when it picks, or is that all fixed in config? we ended up sending the boring lookup subagents to a small model and the bill dropped by about half

白立达 的头像
白立达2 天前

this is the WAY!

Asif Sheikh 💸 的头像
Asif Sheikh 💸2 天前

bring hermes agent modding now please :))

DailyXplorer 的头像
DailyXplorer2 天前

A mode switch waits for the next session by default, because changing the tool list mid-chat would break the prompt cache.

Redmix 的头像
Redmix2 天前

Nice 💪

Giyu 的头像
Giyu2 天前

Difference between having simply 1 orchestator and asking him simply to work with another agent who already exists?

Adria B.A. 的头像
Adria B.A.2 天前

That's awesome and very useful

Otsoa 的头像
Otsoa2 天前

Nice job team !

mekedok 的头像
mekedok2 天前

@Avenoxai bak bakalım işimize yararmı

seridarivus 🇷🇴 的头像
seridarivus 🇷🇴2 天前

Letting the orchestrator choose a different model per subagent route means cost and capability get tuned per task instead of one model doing everything.

Rob Trying Things 的头像
Rob Trying Things2 天前

Oh yes! Perfect for my personal and family fleet of agents!

anya 的头像
anya2 天前

This is a solid upgrade

相关视频

This Hermes update is honestly kind of scary. Your agent now decides which AI to hire for every job. Nous Research just released Advanced Delegation, an official Hermes plugin. When Hermes hands a task to a subagent, it can now pick the right model or profile for it: → A fast, cheap model for lookups, summaries and simple edits → A deep model for the hard thinking → A local model when the data has to stay on your machine → Or one of your Hermes profiles, with its own skills, keys and personality You set the routes. The agent picks one by name. It never sees your keys, so it can't invent a route. One agent. A whole team underneath it. To set it up: → Update to Hermes 0.22.5 or newer → Run: hermes plugins install advanced-delegation → Run: hermes delegation config to add your routes → Type /new to start a fresh chat Want my Hermes setup? DM me "SOP" and I'll send it over 👊
9:24

This Hermes update is honestly kind of scary. Your agent now decides which AI to hire for every job. Nous Research just released Advanced Delegation, an official Hermes plugin. When Hermes hands a task to a subagent, it can now pick the right model or profile for it: → A fast, cheap model for lookups, summaries and simple edits → A deep model for the hard thinking → A local model when the data has to stay on your machine → Or one of your Hermes profiles, with its own skills, keys and personality You set the routes. The agent picks one by name. It never sees your keys, so it can't invent a route. One agent. A whole team underneath it. To set it up: → Update to Hermes 0.22.5 or newer → Run: hermes plugins install advanced-delegation → Run: hermes delegation config to add your routes → Type /new to start a fresh chat Want my Hermes setup? DM me "SOP" and I'll send it over 👊

Julian Goldie SEO

36,030 次观看 • 1 天前

Okay, this one is sneaky good. Hermes can now run every subagent on a different model. Advanced Delegation is an official Nous Research plugin. You set up named model routes, like fast and deep, and Hermes sends each piece of work down the route that fits. ▸ A fast, cheap model for lookups and summaries ▸ Your strongest model for the hard thinking ▸ A local model on your own endpoint ▸ Or another Hermes profile, with its own model, skills and SOULmd Every result shows the route and model it actually ran on. And Hermes only sees the names and descriptions you write, so it can't invent a route. One agent, a whole bench of models. Setup in the reply.
0:26

Okay, this one is sneaky good. Hermes can now run every subagent on a different model. Advanced Delegation is an official Nous Research plugin. You set up named model routes, like fast and deep, and Hermes sends each piece of work down the route that fits. ▸ A fast, cheap model for lookups and summaries ▸ Your strongest model for the hard thinking ▸ A local model on your own endpoint ▸ Or another Hermes profile, with its own model, skills and SOULmd Every result shows the route and model it actually ran on. And Hermes only sees the names and descriptions you write, so it can't invent a route. One agent, a whole bench of models. Setup in the reply.

Hermes Release Watch

26,677 次观看 • 5 天前

my Hermes Agent now sends each subagent to the model that fits the job. I set up the new advanced-delegation plugin from Nous Research with two routes on Nous Portal, then gave it one prompt with two jobs: - a quick summary of Python's heapq went to Claude Haiku 5.5 and came back in 3.0s - a five-pirate logic puzzle went to Claude Opus 5.5 at high reasoning and came back solved in 9.3s the main agent picked the route for each job by itself. the whole run is in the video, setup in the first comment.
1:15

my Hermes Agent now sends each subagent to the model that fits the job. I set up the new advanced-delegation plugin from Nous Research with two routes on Nous Portal, then gave it one prompt with two jobs: - a quick summary of Python's heapq went to Claude Haiku 5.5 and came back in 3.0s - a five-pirate logic puzzle went to Claude Opus 5.5 at high reasoning and came back solved in 9.3s the main agent picked the route for each job by itself. the whole run is in the video, setup in the first comment.

witcheer

13,218 次观看 • 4 天前

Hermes Agent now has an official plugin for MeshyAI. Ask Hermes to make a 70m poly treasure chest for Godot, or to turn a photo of a mug into a model you can print, and it does the job through Meshy's MCP server. In Hermes Desktop the models render in chat, rigged characters animate inline, and a gallery stores every model and prompt. On messaging apps, Hermes sends a short preview video so you can create on the go.
1:11

Hermes Agent now has an official plugin for MeshyAI. Ask Hermes to make a 70m poly treasure chest for Godot, or to turn a photo of a mug into a model you can print, and it does the job through Meshy's MCP server. In Hermes Desktop the models render in chat, rigged characters animate inline, and a gallery stores every model and prompt. On messaging apps, Hermes sends a short preview video so you can create on the go.

Nous Research

371,997 次观看 • 7 天前

Hermes is INSANE as an operating layer for your business. If you’re using it like ChatGPT, you’re probably leaving 90% of the value on the table. The first thing I’d do is create separate profiles by job. Thinking profile: stronger model, strategy, planning, pushback. Execution profile: writing, editing, research, task completion. Workflow profile: one specific loop like CRO, content, sales, SEO, or recruiting. Then I’d give each profile a simple rule: "Don’t just answer me. Rate the idea. Push back. Tell me what’s missing. Help me improve the system." That’s how I’m using Hermes right now:
2:41

Hermes is INSANE as an operating layer for your business. If you’re using it like ChatGPT, you’re probably leaving 90% of the value on the table. The first thing I’d do is create separate profiles by job. Thinking profile: stronger model, strategy, planning, pushback. Execution profile: writing, editing, research, task completion. Workflow profile: one specific loop like CRO, content, sales, SEO, or recruiting. Then I’d give each profile a simple rule: "Don’t just answer me. Rate the idea. Push back. Tell me what’s missing. Help me improve the system." That’s how I’m using Hermes right now:

ericosiu

58,380 次观看 • 3 个月前

Introducing the Hermes Agent Profile Builder You can now build a complete profile in the dashboard with full control over identity/name/description, model/provider, built-in + optional skills, skills-hub installs, and MCP servers in one easy flow
0:47

Introducing the Hermes Agent Profile Builder You can now build a complete profile in the dashboard with full control over identity/name/description, model/provider, built-in + optional skills, skills-hub installs, and MCP servers in one easy flow

Nous Research

546,556 次观看 • 4 个月前

Hermes can now log itself into websites without ever seeing your password. That’s a pretty big upgrade for browser automation. Real Profile already let Hermes use websites you were already signed into. This handles the other case: You’re not logged in yet. Hermes can now securely pull the login from: • Hermes’ encrypted local vault • 1Password • Bitwarden Then fill the credentials and sign in without exposing the raw password to the model or conversation. And it doesn’t stop when 2FA shows up. If your saved login has an authenticator key, Hermes can generate and enter the code automatically. If the site sends you a verification code by text, email, or app, Hermes gives you a secure prompt to enter it there instead of putting it in chat. The code never enters the conversation or model context. Passkeys, hardware keys, and approval taps still stay with you. Hermes waits for you to complete that step, then continues. So browser automation can now go from: “Please sign in” to actually getting through the login and continuing the workflow. Without handing your password or 2FA code directly to the model.
0:23

Hermes can now log itself into websites without ever seeing your password. That’s a pretty big upgrade for browser automation. Real Profile already let Hermes use websites you were already signed into. This handles the other case: You’re not logged in yet. Hermes can now securely pull the login from: • Hermes’ encrypted local vault • 1Password • Bitwarden Then fill the credentials and sign in without exposing the raw password to the model or conversation. And it doesn’t stop when 2FA shows up. If your saved login has an authenticator key, Hermes can generate and enter the code automatically. If the site sends you a verification code by text, email, or app, Hermes gives you a secure prompt to enter it there instead of putting it in chat. The code never enters the conversation or model context. Passkeys, hardware keys, and approval taps still stay with you. Hermes waits for you to complete that step, then continues. So browser automation can now go from: “Please sign in” to actually getting through the login and continuing the workflow. Without handing your password or 2FA code directly to the model.

Hermes Release Watch

45,758 次观看 • 1 个月前

This week's Hermes Agent masterclass is on subagents and delegation! In this clip, I set up a more expensive parent agent for orchestration with a cheaper model for the children subagents doing the research. See how you can get 5x cheaper workflows with this setup!
8:22

This week's Hermes Agent masterclass is on subagents and delegation! In this clip, I set up a more expensive parent agent for orchestration with a cheaper model for the children subagents doing the research. See how you can get 5x cheaper workflows with this setup!

tonbi

43,818 次观看 • 4 个月前

New in the ENS App: set a header image for your ENS name. And if you use the 'Upload Image' option for your header or avatar, you can make gasless changes anytime. Just like your avatar, now you can personalize your profile with a header. Try it now →
0:22

New in the ENS App: set a header image for your ENS name. And if you use the 'Upload Image' option for your header or avatar, you can make gasless changes anytime. Just like your avatar, now you can personalize your profile with a header. Try it now →

ens.eth

53,444 次观看 • 1 年前

Introducing: Agent Profiles AI tools are powerful, but using them is still messy. Different accounts, different providers, different setups. it quickly adds friction. Commander fixes that with Agent Profiles. Each profile is a complete setup for an agent like Claude, Codex, Pi, or OpenCode. Start a chat with any profile, or assign a profile to a project so it always uses the right setup automatically.
1:12

Introducing: Agent Profiles AI tools are powerful, but using them is still messy. Different accounts, different providers, different setups. it quickly adds friction. Commander fixes that with Agent Profiles. Each profile is a complete setup for an agent like Claude, Codex, Pi, or OpenCode. Start a chat with any profile, or assign a profile to a project so it always uses the right setup automatically.

Marcin Krzyzanowski

29,370 次观看 • 7 个月前

Hermes made this quick explainer short on the new subagent controls your orchestrator Hermes Agent can use in v0.21.0 (the voice could be better but eh)
1:03

Hermes made this quick explainer short on the new subagent controls your orchestrator Hermes Agent can use in v0.21.0 (the voice could be better but eh)

Teknium 🪽

30,756 次观看 • 1 个月前

Here are the new destination options in Tesla FSD V14. You can select from 5 parking choices or curbside for Robotaxi-style drop offs. Your preferences for arrival options and preferred parking positions are remembered for each destination. Tesla’s reasoning model also evaluates the best options for your destination and selects the most intuitive default. I’ve experienced all options in my Model Y and they work well. The new speed profile layout is on the top, which includes the new sloth mode. Sloth mode is good for slower or more nervous drivers, or for those that prefer to match the highway speed limit and stay on the right lanes.
0:20

Here are the new destination options in Tesla FSD V14. You can select from 5 parking choices or curbside for Robotaxi-style drop offs. Your preferences for arrival options and preferred parking positions are remembered for each destination. Tesla’s reasoning model also evaluates the best options for your destination and selects the most intuitive default. I’ve experienced all options in my Model Y and they work well. The new speed profile layout is on the top, which includes the new sloth mode. Sloth mode is good for slower or more nervous drivers, or for those that prefer to match the highway speed limit and stay on the right lanes.

Sawyer Merritt

138,692 次观看 • 1 年前

Stop hardcoding one model name in your code. Now you can give each request a policy: a small rule for what the call needs It picks the right model for the job, on your own keys This is unhardcoded, our new open source routing for AI models, live today
0:47

Stop hardcoding one model name in your code. Now you can give each request a policy: a small rule for what the call needs It picks the right model for the job, on your own keys This is unhardcoded, our new open source routing for AI models, live today

GenLayer

98,139 次观看 • 4 个月前

Introducing Bot Mode for Hermes Desktop. Your agent profiles become a series of named Bots. Each Bot has its own role, model, memory, skills and profile picture; Bots can use any model and even communicate with each other. Build a specialist Bot once to use it forever.
1:30

Introducing Bot Mode for Hermes Desktop. Your agent profiles become a series of named Bots. Each Bot has its own role, model, memory, skills and profile picture; Bots can use any model and even communicate with each other. Build a specialist Bot once to use it forever.

Nous Research

2,554,827 次观看 • 2 个月前

today, we're releasing the API for Krea 2. now available in platforms like fal or ComfyUI, through agents like Hermes from Nous Research, and with full support for Claude, Codex, or OpenClaw. learn how you can set it up 👇
0:26

today, we're releasing the API for Krea 2. now available in platforms like fal or ComfyUI, through agents like Hermes from Nous Research, and with full support for Claude, Codex, or OpenClaw. learn how you can set it up 👇

Krea

136,954 次观看 • 5 个月前

factory cli now lets you choose a different model for plan mode and execute mode when you switch modes, the model changes too. my go-to right now: plan: gpt-5-high code: gpt-5-codex (auto)
0:21

factory cli now lets you choose a different model for plan mode and execute mode when you switch modes, the model changes too. my go-to right now: plan: gpt-5-high code: gpt-5-codex (auto)

Ian Nuttall

27,156 次观看 • 1 年前

Kimi track - First place: JumpFoundry by Ev A Hermes Agent plugin where the designer hand-draws one or two glyphs and Hermes Agent generates the rest, acting as coordinator that routes per-glyph revisions through parallel model calls before exporting an editable TTF.
0:41

Kimi track - First place: JumpFoundry by Ev A Hermes Agent plugin where the designer hand-draws one or two glyphs and Hermes Agent generates the rest, acting as coordinator that routes per-glyph revisions through parallel model calls before exporting an editable TTF.

Nous Research

45,952 次观看 • 5 个月前

Sub-agent Model Selection — Different Tasks, Different Models Your main agent runs Qwen3.6-Plus for quality. But not every subtask needs a flagship model. Now sub-agents can use a different model. Create a skill file with model: openai:qwen3.5-plus and the sub-agent runs on that model. Powerful model for the hard parts, fast model for the easy parts. Save tokens without sacrificing quality on what matters.
1:03

Sub-agent Model Selection — Different Tasks, Different Models Your main agent runs Qwen3.6-Plus for quality. But not every subtask needs a flagship model. Now sub-agents can use a different model. Create a skill file with model: openai:qwen3.5-plus and the sub-agent runs on that model. Powerful model for the hard parts, fast model for the easy parts. Save tokens without sacrificing quality on what matters.

Qwen

21,333 次观看 • 6 个月前

Hermes Agent now has an official plugin for Reverse Engineer Anything by morluto ! I made a short video breaking down the plugin and showing how it works hermes plugins install rea That one command will get you set up with the proper MCP config and a special skill your agent can use!
4:26

Hermes Agent now has an official plugin for Reverse Engineer Anything by morluto ! I made a short video breaking down the plugin and showing how it works hermes plugins install rea That one command will get you set up with the proper MCP config and a special skill your agent can use!

tonbi

48,187 次观看 • 13 天前

Instant Claws Every group can now get an exploding one-time OpenClaw🦞 New for every group, isolated to only that convo, no sign up or name required, create a skill or use one from the list Swipe up for a new 🦞 , set the time bomb 💣, invite some friends Invite only for now
2:30

Instant Claws Every group can now get an exploding one-time OpenClaw🦞 New for every group, isolated to only that convo, no sign up or name required, create a skill or use one from the list Swipe up for a new 🦞 , set the time bomb 💣, invite some friends Invite only for now

Shane Mac

20,106 次观看 • 6 个月前