Import ChatGPT and other AI chats: formats and recovery
Bring supported ChatGPT, Claude, Grok or Gab AI conversations into Gab AI. Find the right export format, check imported history and recover from incomplete files.
Enlarge preview
Choose the matching import provider
Select the matching provider and supported export
To import ChatGPT conversations or another supported export, sign in and open Settings → Import. The current file flow offers Gab AI, ChatGPT, Claude and Grok. Select the provider and use supported JSON or ZIP conversation files. A profile or summary index is not necessarily chat history. Provider export options can change; use its current account controls and help.
Format and content checks
| Topic | What to check |
|---|---|
| ChatGPT | A readable conversation message tree is needed. The current parser recognizes supported wrapped records, numbered conversation JSON files and ZIP archives. Summary-only records cannot reconstruct missing messages. |
| Claude | Use a supported export with chat_messages records and readable message content. |
| Grok | Use supported conversation/response structures. Review supported timestamps and actual text. |
| Gab AI | Use supported conversation records from an account export or conversation archive. This does not restore every other account record or setting. |
| Other providers | Do not assume support from a similar filename. Check the current supported flow before converting or uploading. |
Review the file, import and result
Keep the original provider archive. Choose the matching Import action and select the relevant conversation files. Read the parsed source and any warnings; then inspect imported, skipped and error counts. Open a small sample of actual conversations to check titles, dates, message order and useful text before continuing.
Know what text history cannot restore
The parser selects supported visible user/assistant text and the chosen ChatGPT branch. Inactive drafts and internal/tool records can be omitted. Image pointers are not recovered binary attachments. Recognized folder/project metadata can organize supported chats, but missing or unsupported metadata needs separate setup. Memories, model settings and another service’s live tools are separate.
Recover chats from an incomplete export
A summary record can name a conversation without containing its messages. In the import example, supplying the complete messages recovered the missing chat; stable source IDs let the importer skip the records already saved.
Incomplete source, complete source and repeated records
| Topic | What to check |
|---|---|
| Repeat the two parsed chats | 0 new chats, 2 skipped. Stable source IDs matched the stored test records. |
| Supply the missing third record’s complete messages | 3 chats parsed, no parse error; local handler saved 1 and skipped 2 existing records. |
| Unsupported material | image pointers and inactive drafts did not become recovered files or extra messages. |
Resolve source errors before repeating
For summary-only or missing-tree errors, select the full archive or all relevant numbered conversation files. An empty source cannot invent conversations. Review per-record failures and preserve originals. Duplicate checks help retries, but inspect counts and a sample instead of assuming every arbitrary archive imports perfectly.
Continue after importing
Import stores supported history without generating a new answer. Before continuing a conversation, review its contents and any saved context. New AI requests follow the selected model’s access and usage rules.
Bring the conversations you want to continue.
Keep the original export, select the matching provider and inspect the imported conversations. Set up memories and other preferences separately.
Import your chatsSee the migration overview