What's new in HyperNoesis

Verify the new address when an unverified account changes its email

Ask Chat for an image or a video and it hands your request to the Image or Video playground.

It first asks whether to improve your prompt or keep it as it is, then opens the app with the prompt ready.

You can now upscale a clip to 1080p, whether you made it here or uploaded it.

Choose Standard, ready in seconds, or an AI upscale that restores detail in about a minute.

You can now extend any clip from the frame you choose, including videos you upload.

Open a clip's menu, pick Extend and continue the video from that moment; the new part is joined to what came before.

You can now use up to 14 reference images in one video and refer to each one in your prompt.

Insert each picture into your prompt to say what it is for. How many you can add depends on the model.

You can now set a video's first frame, last frame or a loop frame, or use a picture as a reference.

Add the boomerang effect and the clip plays forward, then back to the start. The options depend on the model.

You can now turn on push to talk in a voice call and hold the microphone button while you speak.

Voice now plays sound effects when it tells you a story, and adapts to the way you talk.

Voice now speaks with expression and answers much sooner.

Replies carry tone, small reactions and pauses, and the first words arrive within a second or two instead of several.

In Text to speech you can now add audio tags such as [excited] or [whispers] to your text in one click.

Video playground is here: make short clips from text or a starting frame.

Pick a model such as Veo 3.1, Kling, Seedance or WAN; every clip stays in your history.

Every voice call is now saved as a chat, and you can send photos and text files during the call.

In Chat you can now dictate your message, attach photos for any model that can see them, and start a voice call from the send button.

You can now share your account with family, friends or a team from Account Sharing.

Voice can now see your shared screen or your camera during a call.

You can now sign in with a passkey and protect your account with two-factor authentication.

You can now use HyperNoesis in Greek.

Text to speech is here: turn any text into lifelike speech.

Choose a model and a voice and listen to the result right away.

Chat is here: talk with leading AI models, with reasoning and web search.

Pick a model, ask anything and let it search the web for up-to-date answers.

Knowledge RAG is here: ask questions about your own documents.

Upload a PDF and chat with it; the answers come from what the document says.

Image playground is here: create images with several AI models.

Describe what you want, pick a model and find every image again in your history.

Voice is here: talk with an AI agent in real time.

Speak naturally and the agent answers out loud, with a live transcript of the call.

Your Noë Bot now updates itself only when it is not busy, so an update does not interrupt your work.

Press Update in your dashboard to apply a waiting update right away.

You can now send any file in a chat for your Bot to use on its computer, and long tasks no longer stop halfway.

Each Bot now has a desktop of its own, so several Bots can work on their computers at the same time.

Show me how now tells you what your Bot is doing as it works, can open a site for you, and one OK covers its everyday steps for the whole session.

Noë Bot now speaks Greek: choose Greek from your account menu.

Menus, settings, the Marketplace and messages are in Greek, and your Bots answer and greet you in Greek.

Show me how is here: your Bot points at its screen and walks you through each step.

Ask how to do something and it shows you where to click and explains it out loud. It acts only when you say so.

You can now type passwords, codes and personal details into a secure form that your Bot never sees.

The form is filled straight into the site, so sign-ins and forms work on any site, including government services. Before it submits a form, your Bot can show you a picture of its screen in the chat.

Every Bot can now search the web and run scheduled routines from day one, with nothing to set up.

Every Bot now has an animated Noë Bot avatar that shows what it is doing.

You can now connect apps from the Marketplace in one click, add plugins, and start a Bot from a template.

Apps you connect, such as Google Drive, are offered to every Bot. Bot templates on the Agents tab come with their instructions, skills and apps ready.

You can now create a Bot in one click, talk to several Bots in one chat, and search your chats.

Noë Bot now runs on a dedicated server of your own, with 4, 8, 12 or 24 GB of memory.

Your Bots, their files and their computers live on a server that is yours alone.

You can now open your Bot's screen full size, take control, and record your steps to teach it a workflow.

Noë Bot is here: AI teammates with their own computer, browser, files and tools.

Every subscriber gets a Noë Bot of their own. Watch your Bots work on their computer's desktop and step in when they need you.

Automatic clean-up of old chats

Hermes Agent can now do new things

Your Hermes server now updates itself only when it is not busy, and only at night if you use your own key.

With your own key, updates happen between 03:00 and 05:00 your time. Press Update in your dashboard to apply a waiting update right away.

Hermes is here: your own always-on Hermes Agent on a dedicated server.

Talk to it on Telegram, Discord or WhatsApp, link WhatsApp by scanning a code, and approve who may message it. Choose 4, 8, 12 or 24 GB of memory.

Your OpenClaw server now updates itself only when it is not busy, and only at night if you use your own key.

With your own key, updates happen between 03:00 and 05:00 your time. Press Update in your dashboard to apply a waiting update right away.

OpenClaw now runs version 2026.10.7.

OpenClaw now runs on a dedicated server of your own, with 4, 8, 12 or 24 GB of memory.

You can now talk to your OpenClaw on WhatsApp and Discord too, and link WhatsApp by scanning a code.

OpenClaw now runs version 2026.9.8 with a working browser and more built-in skills, and you can update it from your dashboard.

OpenClaw is here: your own always-on OpenClaw agent on Telegram.

Pick a model, use your own key or your HyperNoesis credits, and manage it from your dashboard: start and stop it, approve who may talk to it and read its logs.