---
title: "AlphaSignal Llms: Translation News"
url: "https://alphasignal.ai/topic/llms/translation"
description: "Latest AI and machine learning news about Translation (Llms) on AlphaSignal."
---

# AlphaSignal Llms: Translation News

Latest AI and machine learning news about Translation (Llms) on AlphaSignal.

## Articles

- [Tencent's Hy-MT2 Brings Offline AI Translation to iPhone in 33 Languages](https://alphasignal.ai/news/tencent-s-hy-mt2-brings-offline-ai-translation-to-iphone-in-33-languages.md) (2026-09-24)
- [Cohere Labs' North Small Translate Beats DeepL and Google Translate Across 50 Languages](https://alphasignal.ai/news/cohere-labs-north-small-translate-beats-deepl-and-google-translate-across-50.md) (2026-09-10)
- [McGill NLP's AfriqueQwen3.5 Beats Larger Models Across 50 African Languages](https://alphasignal.ai/news/mcgill-nlp-s-afriqueqwen3-5-beats-larger-models-across-50-african-languages.md) (2026-08-30)
- [Sakana AI's Namazu Beats Google Translate and DeepL on Japanese Cultural Nuance](https://alphasignal.ai/news/sakana-ai-s-namazu-beats-google-translate-and-deepl-on-japanese-cultural-nuance.md) (2026-08-20)
- [Sarvam AI's Indus Wins Maharashtra, Putting Sovereign AI to Work for 2,500 Officials](https://alphasignal.ai/news/sarvam-ai-s-indus-wins-maharashtra-putting-sovereign-ai-to-work-for-2-500.md) (2026-08-12)
- [Cohere Labs' Aya-23-8B Beats Mistral and Gemma Across 23 Languages](https://alphasignal.ai/news/cohere-labs-aya-23-8b-beats-mistral-and-gemma-across-23-languages.md) (2026-07-09)
- [Cohere Labs' Tiny Aya Brings 70-Language AI to Laptops Without Internet](https://alphasignal.ai/news/cohere-labs-tiny-aya-brings-70-language-ai-to-laptops-without-internet.md) (2026-07-09)
- [Sakana AI's Namazu Tackles What Google Translate Gets Wrong About Japanese Tone](https://alphasignal.ai/news/sakana-ai-s-namazu-tackles-what-google-translate-gets-wrong-about-japanese-tone.md) (2026-07-06)
- [ETH Zurich Trains AI to Read Grammar Books and Translate 7,000 Languages](https://alphasignal.ai/news/eth-zurich-trains-ai-to-read-grammar-books-and-translate-7-000-languages.md) (2026-06-04)
- [VISTEC's DuDi Teaches Tiny AI Models to Speak Southeast Asian Languages](https://alphasignal.ai/news/vistec-s-dudi-teaches-tiny-ai-models-to-speak-southeast-asian-languages.md) (2026-06-03)
- [University of Bucharest Boosts Romanian Coreference Resolution by 19 Points Without Training Data](https://alphasignal.ai/news/university-of-bucharest-boosts-romanian-coreference-resolution-by-19-points.md) (2026-06-03)
- [Cohere's Command A+ Beats Google Translate on 48 Languages Running on 2 GPUs](https://alphasignal.ai/news/cohere-s-command-a-beats-google-translate-on-48-languages-running-on-2-gpus.md) (2026-05-29)
- [Tencent's Hy-MT2-7B Beats DeepSeek on Translation and Runs Locally](https://alphasignal.ai/news/tencent-s-hy-mt2-7b-beats-deepseek-on-translation-and-runs-locally.md) (2026-05-29)
- [Tencent's Hy-MT2 Beats Commercial Translation APIs at 440MB on Device](https://alphasignal.ai/news/tencent-s-hy-mt2-beats-commercial-translation-apis-at-440mb-on-device.md) (2026-05-26)
- [Tencent's Hy-MT2 Beats DeepSeek at Translation and Fits in 440MB](https://alphasignal.ai/news/tencent-s-hy-mt2-beats-deepseek-at-translation-and-fits-in-440mb.md) (2026-05-21)
- [Tencent's Hy-MT2 Squeezes 33-Language Translation Into 440MB on Your Phone](https://alphasignal.ai/news/tencent-s-hy-mt2-squeezes-33-language-translation-into-440mb-on-your-phone.md) (2026-05-11)
- [Tencent's Hy-MT1.5 Beats 72B Models Squeezed Into 440MB on Your Phone](https://alphasignal.ai/news/tencent-s-hy-mt1-5-beats-72b-models-squeezed-into-440mb-on-your-phone.md) (2026-04-28)
- [Beijing Academy of AI's LaoBench Exposes Massive Gaps in Multilingual AI](https://alphasignal.ai/news/beijing-academy-of-ai-s-laobench-exposes-massive-gaps-in-multilingual-ai.md) (2026-04-15)
- [A*STAR's AdaMCoT Lets AI Think in English and Answer in Any Language](https://alphasignal.ai/news/a-star-s-adamcot-lets-ai-think-in-english-and-answer-in-any-language.md) (2026-04-14)
- [Tübingen AI Center's LiT Benchmark Exposes How Multilingual AI Tests Measure English Math](https://alphasignal.ai/news/tubingen-ai-center-s-lit-benchmark-exposes-how-multilingual-ai-tests-measure.md) (2026-04-14)
- [UC Santa Barbara Proves English-Only Fine-Tuning Hurts Even English AI Accuracy](https://alphasignal.ai/news/uc-santa-barbara-proves-english-only-fine-tuning-hurts-even-english-ai-accuracy.md) (2026-04-14)
- [French Linguists Release GLeMM, a 2.3M-Pair Open Word-Meaning Lexicon](https://alphasignal.ai/news/french-linguists-release-glemm-a-2-3m-pair-open-word-meaning-lexicon.md) (2026-04-14)
- [TAPS Forces Diffusion LLMs Like LLaDA to Stop Repeating Themselves](https://alphasignal.ai/news/taps-forces-diffusion-llms-like-llada-to-stop-repeating-themselves.md) (2026-03-16)
- [Kathmandu University Builds NepTam to Finally Put 1.4M Speakers on the AI Map](https://alphasignal.ai/news/kathmandu-university-builds-neptam-to-finally-put-1-4m-speakers-on-the-ai-map.md) (2026-03-14)
- [Cohere Labs' Tiny Aya Beats Gemma on 46 Languages at 3B Parameters](https://alphasignal.ai/news/cohere-labs-tiny-aya-beats-gemma-on-46-languages-at-3b-parameters.md) (2026-03-12)
- [Brown and Berkeley Expose How Multilingual AI Benchmarks Hide 30% Performance Gaps](https://alphasignal.ai/news/brown-and-berkeley-expose-how-multilingual-ai-benchmarks-hide-30-performance.md) (2026-03-12)
- [Saarland Researcher Explains Why Translated Text Always Feels Slightly Off](https://alphasignal.ai/news/saarland-researcher-explains-why-translated-text-always-feels-slightly-off.md) (2026-03-12)
- [Every AI Model Tested Defaults to Masculine When Translating Basque, Study Finds](https://alphasignal.ai/news/every-ai-model-tested-defaults-to-masculine-when-translating-basque-study-finds.md) (2026-03-09)
- [Google Releases TranslateGemma to Rival Closed APIs Across 55 Languages](https://alphasignal.ai/news/google-releases-translategemma-to-rival-closed-apis-across-55-languages.md) (2026-01-12)
- [Google's Flan-T5-XXL Beats 62B Models at a Fraction of the Size](https://alphasignal.ai/news/google-s-flan-t5-xxl-beats-62b-models-at-a-fraction-of-the-size.md) (2022-10-21)
- [Google's FLAN-T5 Proves a 250M Model Can Beat Giants on Zero-Shot Tasks](https://alphasignal.ai/news/google-s-flan-t5-proves-a-250m-model-can-beat-giants-on-zero-shot-tasks.md) (2022-10-21)
- [Meta's NLLB-200 Translates 200 Languages With 44% Better Accuracy](https://alphasignal.ai/news/meta-s-nllb-200-translates-200-languages-with-44-better-accuracy.md) (2022-07-08)
- [BigScience's BLOOM Covers 46 Languages Where Most Open Models Fail](https://alphasignal.ai/news/bigscience-s-bloom-covers-46-languages-where-most-open-models-fail.md) (2022-05-19)
- [Hugging Face and Microsoft Turn Azure Model Deployment Into a Single Click](https://alphasignal.ai/news/hugging-face-and-microsoft-turn-azure-model-deployment-into-a-single-click.md) (2022-03-02)
- [Helsinki Refreshes opus-mt-fr-en With 1M Monthly Downloads and a Breaking Change](https://alphasignal.ai/news/helsinki-refreshes-opus-mt-fr-en-with-1m-monthly-downloads-and-a-breaking-change.md) (2022-03-02)
- [Helsinki-NLP's MultiSynt/MT Gives 36 Languages 4.8T Tokens Using Tiny Translation Models](https://alphasignal.ai/news/helsinki-nlp-s-multisynt-mt-gives-36-languages-4-8t-tokens-using-tiny.md) (2022-03-02)