v1.2.1

• TgCrypto fix: Added full support for modern Python versions. The issue with being unable to compile the original tgcrypto on Python 3.12/3.13+ has been resolved by switching to a compatible fork that builds reliably on Python 3.8 – 3.14.x.
• Fixed issue #3: Fixed connection leak. Temporary clients created during CreateNewSession (QR login) are now guaranteed to disconnect and close the session after extracting the data. (special thanks to LmingXie for the report)
• Fixed issue #4 : Updated .tl schema files. Fixed connection errors to new sessions ("Unknown key type in encrypted map") (special thanks to dxccry for the report).
• Lazy Loading: Parsing of device databases (_LazyData) now happens lazily on first access. The library no longer hangs the script on import/startup.
• Auto-Update: Completely rewritten version fetcher. Now it parses not only the latest app versions, but also build codes directly from the official channel t.me/s/tgstable and Homebrew. The library always uses the most up-to-date version + build combination, allowing it to mimic a real device with top-tier fingerprints as accurately as possible.
• MTProto Layer: Updated the base schema to 225. Added smart automatic synchronization with installed Telethon version. The library dynamically detects its layer (telethon.tl.alltlobjects.LAYER) and uses the highest one.
• Docs: Documentation moved to GitHub Pages.
  Suggested changelog entry (ready to paste, in the exact style):

v1.1.6

• Fixed issue #1 (special thanks to Mekotofeuka for the report) — added Python 3.13+ support
• Fixed lang_pack empty value error occurring in newer Telethon versions

v1.1.5

• Updated versions of official applications and added an automatic fetcher that obtains the latest current versions
• Updated version of Telegram API Layer
• Post login steps to simulate getConfig, nearestDc, getLanguages, and other calls that occur in the official application
• Fingerprints database updated, obsolete junk devices removed, weight parameter and min, max SDK versions added so that everything is generated more statistically realistically, without anomalies

v1.1.0

• Fingerprints & Consistency: Added a system for generating realistic device fingerprints and the ConsistencyChecker tool. This allows you to mimic the behavior of official clients and check session parameters for compliance with Telegram standards to reduce the risk of blocking.
• Support for .session + .json: Import and export of accounts in .session (SQLite) + .json (API metadata) formats has been implemented. This allows for easy transfer of sessions between different tools while preserving all hardware settings.
• Web User-Agents: Integration with the browserforge library has been added to generate up-to-date browser User-Agents. Now the web versions (Z, A, K, and Webogram) use realistic fingerprints of modern browsers.

v1.0.0

• The PyQt5 library has been removed. Its functionality has been reimplemented in pure Python to eliminate overhead and avoid installing unnecessary heavy dependencies.
• Added Web Telegram A version support
• Some improvements of device information generation. The library now uses a new device database.