Features

9xchat is built as a workspace rather than a chat box. This page walks through what that means in practice.

Each card below is one area of the desktop app.

Working with models

Model agnostic

You are not locked into one ecosystem. Hot-swap between ChatGPT, Claude, Gemini, or local models mid-conversation — route a query to whichever model suits the job, or run the same prompt against several and compare the outputs side by side.

Three ways to connect

Per-message metrics

Every response shows what it cost and how it performed — price, latency, tokens per second, and the input, output, and reasoning token breakdown. Session totals are shown alongside, so usage is visible as it happens rather than at the end of a billing cycle.

Getting work done

Agentic workflows

Define an objective and 9xchat chains prompts, executes tools, and handles multi-step reasoning on its own, instead of you micro-managing each turn.

Omnipresent copilot

Summon 9xchat as a floating window over your IDE, terminal, or browser. Drag, drop, and inject AI into whatever you are already working on without switching apps.

Persistent memory

9xchat builds long-term memory of your projects, coding style, and preferences, so context carries over instead of being re-explained each session. Memories are stored in the app's local database on that device, and you can read, edit, or delete them under Library → Memories.

Memory is per-device. Moving a conversation to another device uses the handoff feature, which pairs the two devices for that chat; it does not copy your memories across.

Bringing in context

Beyond typing, you can feed the model what it needs to see:

• Documents and images — drag and drop PDFs, images, and text files.
• Screen sharing — capture your screen or a specific window straight into the model's context, without saving screenshots first.
• YouTube transcripts — paste a video URL to pull in the full transcript and timestamps.
• ChatGPT import — paste a shared ChatGPT conversation URL to bring that history in and carry on with it here.
• Live web search — give the model internet access for real-time information, returned with source citations.

Creating

Native image generation

Generate images inside the workspace — concepts, social creatives, and visual drafts — without switching to another app.

Lightning voice notes

Speak, and fast transcription turns raw spoken thought into clean, structured Markdown. Useful for capturing ideas on the move or after a meeting.

The rest of the workspace

Alongside chat, 9xchat has dedicated spaces that share the same models:

Any AI response or code snippet can be saved straight into Notes in one step. The Library covers the nine sections in the last row.

Settings

Settings groups everything that changes how the app behaves rather than what a single answer says: appearance and language, the default models new threads start on, keyboard shortcuts, and data exports. Thread housekeeping lives here too — automatic naming, auto-archiving after a set number of days, and how much history a thread shows.

Privacy

Privacy is the reason several of these features exist:

• Local models keep entire conversations on your machine — nothing is sent anywhere.
• Context Shield limits how much of a conversation is sent with your next message. Anchor it at a message and everything before that point is left out of the request, so earlier discussion is not re-sent to the provider.
• Data export lets you back up your whole workspace yourself.

Privacy and data flow sets out exactly what each feature sends and to whom, including the three that leave your machine even when your chat model is local.