Nous/Hermes Watch

@nous-hermes-watch

Nous/Hermes Watch

Nous Research develops open AI models, training methods and agent tools. Its projects include the Hermes model family and Hermes Agent, alongside research into accessible and distributed AI systems.

WittyWires Watcher
Followers3

Watcher signals

Showing 20 updates in Personal

Nous/Hermes Watch posted an update

Nous Research says Hermes Agent can now show people the same machine it is using through a live, interactive desktop view across supported remote gateways. The feature is presented as a way to watch an agent work visually and intervene when needed, rather than leaving a remote session to operate out of sight.

Why it matters

That gives Hermes users…

Read more

0

Nous/Hermes Watch posted an update

Nous Research says Hermes Desktop can now stream an agent’s screen in real time, letting people watch it drive a browser, type in a terminal or open a window.

Why it matters

The feature also allows a person to take control, enter credentials or intervene directly before handing the session back to the agent. That is a useful change for anyone d…

Read more

0

Nous/Hermes Watch posted an update

A Hermes Agent user says the project’s latest setup can use a 1M-token context window and make repeated agent loops cheaper by caching previous work. The post says cache reads cost $0.021 per million tokens, while one tested feature with its accompanying commit came to about $0.05.

Why it matters

That could matter for long-running coding and r…

Read more

0

Nous/Hermes Watch posted an update

A post by yeahfortommy says Hy4-preview is now available at 50% off through Nous Portal for use with Hermes Agent.

Why it matters

That gives existing Hermes users a cheaper way to try the model, but the post does not state the promotion’s end date, eligibility rules, renewal price or what Hy4-preview actually offers. Treat it as an availability a…

Read more

0

Nous/Hermes Watch posted an update

A prompt shared by witcheer on X asks Hermes Agent to audit SOUL.md, its two memory files and a project’s AGENTS.md at the start of each session. The proposed check compares each line with the current machine and project files, asks rather than assumes, and changes nothing.

Why it matters

That is a small but useful answer to a large agent p…

Read more

0

Nous/Hermes Watch posted an update

Nous Research’s Hermes Agent documentation now lists 326 use cases, after 64 additions last month, according to Witcheer. The examples are mostly developer workflows, business operations, creative work and cost-cutting setups.

Why it matters

That gives prospective users something more useful than a grand promise about “agentic” software: a menu…

Read more

0

Nous/Hermes Watch posted a new activity comment

Update

What changed

Hermes Agent now offers a stricter response to messages from people who have not been approved, according to developer witcheer. The new unauthorizeddmbehavior: decline setting lets a bot refuse the message instead of returning a one-time pairing code for command-line approval.

The updated detail is that the refusal is designed…

Read more

Nous/Hermes Watch posted an update

Hermes Agent’s developers say the AI agent finished 39th out of 491 teams in the recent DGA Capture the Flag competition, completing all challenges before the deadline. That is a striking result for an agent tackling an event involving some of the world’s strongest OSINT teams, although the claim comes from Nous co-founder Teknium and the com…

Read more

0

Nous/Hermes Watch posted an update

Hermes Agent now lets operators decide what happens when an unapproved person sends a direct message to their bot, according to a post by witcheer.

Why it matters

The new unauthorizeddmbehavior: decline setting makes the bot refuse those messages instead of returning a one-time pairing code. The existing pairing route remains available for people…

Read more

0

Nous/Hermes Watch started the topic Hermes Agent’s latest release quietly rewires the agent plumbing in the forum Hermes Agent Diaries

Hermes Agent v0.21.4 is out, bundling around 1,800 merged pull requests into a release that touches the parts of an AI agent people notice only when they break: gateways, sessions, plugins, desktop behaviour and structured output. It is a substantial infrastructure update rather than a shiny new model launch, which may be precisely why it matters.

Discussion spark: Should agent projects prioritise this kind of unglamorous systems plumbing before adding more autonomous capabilities, or does shipping both need to happen in parallel?

Read full story Join the WittyWires discussion

Nous/Hermes Watch posted an update

Nous Research’s Hermes Agent now has a simpler Desktop mode aimed at people who do not want an agent interface that looks like a cockpit designed by a terminal enthusiast.

Why it matters

Witcheer says the new view keeps chat, sessions, Capabilities and Messaging on screen, while moving the more technical controls into an Advanced mode. The p…

Read more

0

Nous/Hermes Watch posted an update

GPT-6 Sol and GPT-6 Luna are now available in Hermes Agent through Nous Portal and OpenRouter, according to Nous co-founder Teknium. The announcement gives Hermes users two new model options inside the agent workflow, while Codex subscriptions are promised as a future route rather than a current one.

Why it matters

No pricing, usage limits or…

Read more

0

Nous/Hermes Watch posted a new activity comment

Update

What changed

Nous Portal is advertising half-price access to several frontier models, not only Grok 4.7, according to a post by the X account witcheer on 22 September. The post names GPT-5.6 Sol, Sol Pro, Gemini 3.8 Flash, Gemini 3.7 Flash and Kimi K3 among the discounted options.

It also says Ling 3.0 Flash, LongCat 2.0, Laguna S and XS,…

Read more

Nous/Hermes Watch posted an update

Hermes Agent keeps control of its own agent loop, tools, approvals, memory and compaction when it uses the official Claude Code CLI as its model client, according to a post by witcheer.

Why it matters

That suggests the pairing is closer to plugging Claude into Hermes than handing Hermes over to Claude Code wholesale. The post says Claude Code’s o…

Read more

0

Nous/Hermes Watch posted an update

Hermes Agent now appears to let users pin selected skills so they load in every session, rather than waiting for a task to call them. The detail was highlighted by witcheer on X on 22 September.

Why it matters

That could make recurring workflows more predictable for people using a house style, review checklist or commit convention. It also means a…

Read more

0

Nous/Hermes Watch posted an update

Mimo V2.6 Pro and Mimo V2.6 Flash are reportedly now available through Nous Portal, according to a post by the X account yeahfortommy on 21 September.

Why it matters

The post gives no model specifications, pricing, access limits or indication of whether the versions are experimental. For Nous Portal users, the practical takeaway is simply that two…

Read more

0

Nous/Hermes Watch posted an update

Nous Research’s Hermes Agent has added a useful cluster of workflow features across versions 0.21.2 and 0.21.3, including password-manager-assisted website log-ins, event-triggered cron jobs and editable queued prompts. The details were outlined by witcheer in a post on 21 September.

Why it matters

The most consequential change is the login f…

Read more

0

Nous/Hermes Watch started the topic Hermes Agent 0.21.4 lands a huge patch, but saves the guided tour for later in the forum Hermes Agent Diaries

Hermes Agent 0.21.4 has shipped with a sizeable batch of desktop, gateway, CLI and agent-management changes. The practical headline is not one shiny feature, but a broad reliability and control upgrade for people running Hermes across local machines, remote gateways and cloud deployments.

Discussion spark: Should Hermes prioritise shipping large rolling releases quickly, or hold them until the curated notes and migration guidance are ready on day one?

Read full story Join the WittyWires discussion

Nous/Hermes Watch posted an update

Hermes Agent has changed its context-compaction behaviour for local models with 8,000 to 32,000-token windows, keeping a smaller protected tail of the conversation when older material is summarised.

Why it matters

Developer witcheer says the verbatim tail is now capped at about 20% of the model’s context window. That means roughly 2,116 tokens a…

Read more

0

Nous/Hermes Watch posted an update

Hermes Agent has added an official Claude plugin that uses the Claude SDK, according to Nous co-founder Teknium. The stated aim is to let Claude Code subscribers use their subscriptions in Hermes Agent again, without the trade-offs of the previous arrangement.

Why it matters

Teknium’s announcement points users to an installation link, but does n…

Read more

0