Importing Chats

Importing lets you bring your existing ChatGPT conversations into just4o.chat so you can keep working without starting over. Your exported chats appear in the normal chat list with full message history, context, and branching preserved. You keep your favorite exchanges and gain access to multiple AI providers (OpenAI, Anthropic, Google, Grok) from a single place.

Full importing requires the Aurora plan. The migration tool itself is free. You can parse exports, preview conversations, download chats as files, and create personas without a subscription. Aurora subscribers can bring up to 100 conversations into their account. Each imported chat keeps its original title and message order. You can open any conversation and continue the dialogue with the same model family or a different one.

Request a data export from ChatGPT via Settings, Data controls, Export data. Once the export arrives (usually by email within 24 hours), locate the conversations.json file inside the downloaded zip.

Go to the migration tool at just4o.chat/migrate. Drag and drop your conversations.json onto the upload area. We parse it and show your last 100 conversations (or up to 500 for picker selection). You can preview any conversation, rename titles, and edit messages before importing. Choose which chats to bring in, then click Import selected. Aurora subscribers can import up to 100 conversations.

Imported chats appear in your chat sidebar with full threads intact. Each retains its original title and message order. Regenerated branches and edited messages are flattened into a single chronological thread during import. You can open any conversation and continue the dialogue with the same model or a different one.

When importing, you can opt in to Style RAG for selected chats. This indexes excerpts from your imported conversations so the model can match your preferred tone and style in brand new chats where the thread has no history yet.

Enable it per chat with the star icon in the import picker, or turn on Auto embed migrated chats to embed your most recent 10 by default. The system embeds user and assistant pairs, then at runtime retrieves the most relevant exemplars by semantic similarity and injects them as stylistic guidance. The model uses them for tone, brevity, and formatting only, not as factual memory.

Your words stay yours. The system instructs the model not to copy names, numbers, or claims unless you provide them in the current chat, and not to quote exemplars verbatim unless you ask. The result is a familiar tone on first move, even in a new chat, without twisting your words. Style RAG applies only to global (non-project) chats and is skipped when a persona is active.