Lightling - web page translator version history - 4 versions
Lightling - web page translator by Hewel
Lightling - web page translator version history - 4 versions
Be careful with old versions! These versions are displayed for testing and reference purposes.You should always use the latest version of an add-on.
Latest version
Version 7.3.0
Released Aug 25, 2026 - 8.88 MBWorks with firefox 58.0 and later- Page translation now prioritizes what you're reading. A new visibility-lane scheduler continuously reranks pending translation units by live viewport position instead of their static scan order, and sizes request batches per lane — visible content is translated and displayed first, even on pages that shift as you scroll.
- Selected-text translations keep their formatting. Rich text selections (bold, links, structure) are now rendered with the source's formatting in the translation popup, and the popup can be dragged anywhere on the page (opt-in setting).
- Google AI Studio (Gemini) support and smarter LLM page translation. A ready-made Gemini preset joins the LLM providers, and page-translation parallelism/batching now adapts automatically to your model's size tier instead of using one fixed concurrency for everything.
- Far fewer wasted requests on dynamic sites. The page translator no longer restarts on query-only URL changes, backs off from volatile page regions that fight the translator, and recognizes proper names/emails as text that legitimately stays untranslated.
- Page translator: visibility-lane priority scheduling. Undispatched translation units are reranked by live viewport lanes, with per-lane batch sizing, so on-screen content is translated first. (b5615109, ace85619, b7413799)
- LLM: Google AI Studio preset with Gemini models. New "Google AI Studio" provider preset (default model
gemini-3.7-flash, withgemini-3.5-flash-litealso registered) usable out of the box for text and page translation. (7215513e) - Page translator: size-tiered adaptive parallelism. Translation concurrency and batch token budgets are chosen from the model's size tier (small/medium/large, inferred from registered profiles, model name, or pricing) and are continuously calibrated by an adaptive budget controller — small models run many parallel requests, large models stay conservative. (d39aaa42, 368c0ac2, ADR 28e22597)
- Select translator: rich-formatted translations. Selected text with HTML formatting is translated via the placeholder-markup protocol and rebuilt as sanitized, semantic markup in the popup. (2d129e31, 1d967766)
- Select translator: draggable popup. New option "Allow dragging the translation popup" (off by default) lets you grab the popup on its background and move it anywhere on the page. (291e5d70)
- Virtualized language selector. The language panels now use a new virtualized Selector component with search, keeping long language lists fast. (ef77ad2f)
- Page translator: semantic unit liveness validation and rescan stabilization. Translation units are tracked for liveness and stale units are never applied; rescans are debounced (~400 ms) so rapid DOM churn no longer causes duplicate work or flicker. (0a2d8d4a)
- Page translator: no more full re-translation on tab-style routers. Sites that switch tabs by rewriting only the U
Source code released under BSD 2-Clause "Simplified" License
Older versions
Version 7.2.0
Released Aug 24, 2026 - 8.82 MBWorks with firefox 58.0 and laterSource code released under BSD 2-Clause "Simplified" License
Version 7.1.65535.1
Released Aug 24, 2026 - 8.82 MBWorks with firefox 58.0 and laterSource code released under BSD 2-Clause "Simplified" License
Version 7.1.0.2
Released Aug 20, 2026 - 4.34 MBWorks with firefox 58.0 and laterSource code released under BSD 2-Clause "Simplified" License