What’s new in
Vitra Universe.

New VitraTM memories follow built-in rules for 1,175 target languages, including numbers and plurals.

• Used in AI translation, the review pass and QE reports
• Your own style guide rules win where they disagree
• Memories created earlier keep translating as before

AI translation uses more of what you already know.

• Each text can carry its own context, such as an i18n key
• The model sees approved memory wording for phrases inside each text
• Long documents keep the same wording throughout
• Terms now match in Chinese, Japanese, Korean and Thai

• One voice library across Playground and Translate Video
• Freed clones are re-created automatically
• 6,400+ new native voices

• Subtitles on two lines of 42 characters
• Split view scrolls both languages together
• Sharper exports

• Quick Create: chat groups with shared rules
• Canva: sign in with your email and stay connected
• One language order in every picker
• QE reports cost 4 credits per source word

• The QE Unverified only toggle works
• Audio-only exports with lip-synced cards work on retry
• 480p and 360p exports have no black bars
• Adding a language no longer overwrites earlier ones

Keep names verbatim and wordings fixed, beside term bases and style guides.

• Mark entries do-not-translate or case-sensitive
• Import and export CSV and Excel
• Start a style guide without files

Score a translation with MQM in Translate Video, Translate Photo and Playground.

• Errors by severity, with corrected text per segment
• Apply the suggested fixes from the report
• Download the report as a PDF

• Brand Kit has its own page in the nav rail

• Playground: translate several files at once, with live progress
• Prompts and memories keep every version
• Translate Photo: send Quick Create images straight to translation
• Dashboard is now Home, and Translation Memory is now TMS

• Running out of credits in the video editor explains the shortfall
• The memory picker lists your memories
• Re-importing a term base no longer duplicates concepts
• Re-imported TM segments keep their metadata

AI translation now follows your style guides and terminology.

• Upload up to 10 files and the rules are extracted for you
• Approved terms are binding, forbidden ones never used
• Import and export TMX, XLIFF, XLSX, CSV and TSV
• VitraTM covers all 1,175 languages in the catalogue

• A generic LLM node: pick the model, modalities and instructions
• Text and Office inputs can feed image and video models
• Preview any step with View output
• Refreshed GPT, Gemini and image model catalogue

• Choose your lip-sync model, with its price shown on export
• OpenAI is the default for new image translation and Image Creator users
• Office translation handles CSV and XLSX
• Organization API keys for agents and scripts

• Bengali and Hindi terms are matched
• Languages that shared a name resolve to the right code
• Lip-sync exports use the organization's chosen model
• Large Gemini batch translations no longer fail after payment