Account controls
Set defaults here: custom instructions, profile details, personas, response font, image settings, and voice preferences.
Chat
Pick the model. Keep the context. Finish the work.
just4o.chat turns chat into a real operating surface for serious AI work: model choice stays explicit, projects keep files and memory in scope, canvas opens beside the thread, and voice plus web search stay right there when you need them.
No model rouletteProjects keep chats, files, and memory togetherCanvas opens beside the threadVoice and search stay in reach
One operating surface
ModelMemoryCanvasVoice
Operating model
The stack stays legible, even when the work gets deep.
That is part of the product advantage. Account sets the baseline, chat runs the active session, and projects hold long-running work together without smearing context across everything else.
- 1
Set your baseline once.
Save instructions, tone, profile details, response font, personas, and voice settings so new chats do not start from scratch.
- 2
Choose the active chat setup.
Pick the model, attach a persona, turn tools on, and decide whether the work belongs in your global stream or inside a project.
- 3
Work from the same surface.
Search the web, open canvas, attach files, use voice, or keep drafting without hopping across separate products.
Chat controls
Choose the model, turn on web search or image tools, open canvas, talk by voice, and keep the current conversation moving.
Project controls
Add a project when the work needs its own files, chats, memory, and a dedicated workspace instead of living in your global stream.
Capabilities
What the chat surface actually includes.
This is not chat for chat's sake. It is a working surface built to keep model choice, memory, tools, and output locked together while you actually ship.
Customization
Your defaults live in Account, not in a saved prompt doc.
Set custom instructions, nickname, occupation, more about you, tone controls, response font, and global files once, then let new chats start from there.
Adaptive memory
Memory is scoped, reviewable, and tied to the active workspace.
Global, project, and persona memory stay distinct. Foreground read and write tools can retrieve or update memory in the current scope, and the UI surfaces memory update badges when something changes.
Personas
A persona is a reusable behavior layer, not a one-off gimmick.
Create personas in Account, attach them to chats or projects, and keep their voice and memory consistent even when you switch models.
Projects
Projects keep chats, files, and memory in one working space.
Each project has its own chat list, files, icon, goal, and memory layer so long-running work stays coherent instead of leaking into everything else.
Canvas
Canvas turns chat into an actual drafting surface.
Open a side-by-side document, edit files in place, review proposed hunks, and keep the conversation attached to the artifact you are producing.
Voice
Voice input and read-aloud stay inside the same chat flow.
Studio and above unlock voice input and read-aloud, while voice model, voice style, silence timing, and playback controls live in Account settings.
Tools
Search, image generation, data, voice, and canvas stay in one picker.
The tools button in the composer controls live web search, generated images, EIA plus BLS data, voice mode, and canvas without sending you to a different surface.
Import
You can start from history instead of rebuilding context from zero.
The migration tool parses ChatGPT exports, helps separate chats into files, and on Aurora can import up to your latest 100 conversations directly into chat.
Providers
The chat stack spans every provider we ship.
just4o.chat is not locked to one lab. The composer can route across OpenAI, Claude, Gemini, Grok, DeepSeek, Qwen, Moonshot, MiniMax, Z.ai, and our own Echo models without making you leave the same workspace.
OpenAI / ChatGPT
GPT-4o, GPT-5.x, GPT OSS, and GPT audio mini live in the same chat surface.
Claude
Opus, Sonnet, and Haiku are all available when you want Anthropic behavior.
Gemini
Gemini 2.5 and Gemini 3 preview families are available for multimodal and long-context work.
Grok
Grok 3, Grok 4, Grok 4.1, and Grok Code are all in rotation here.
DeepSeek
DeepSeek chat models are available for efficient long-context reasoning.
Qwen
Qwen 3.6 Plus is available for flagship multimodal chat with function calling and image input.
Moonshot
Kimi models are in chat for reasoning-heavy and multimodal workflows.
MiniMax
MiniMax models are available for agentic, multilingual, and office-style work.
Z.ai
GLM families are available for coding, systems work, and high-context chats.
just4o.chat
Echo and Echo mini give the app its own house voice inside the same workspace.
Limits and guardrails
The chat product is opinionated about access instead of hiding it.
just4o.chat exposes when a feature is gated, limited, or unavailable. The UI would rather tell you the truth than silently switch behavior in the background.
Explorer gets 20 user messages per chat, Builder 250, Studio 500, and Aurora plus Pro remove the chat-length ceiling.
Explorer has no web search. Builder gets 5 per day, Studio 10, Aurora 25, and Pro 50.
Voice runs on GPT audio mini in chat, with GPT-4o mini TTS, Grok TTS, and ElevenLabs Flash v2.5 available for read-aloud.
Start where the product actually lives.
The chat surface is where your model choice, memory, files, voice, and canvas meet.