#multilingualai — Public Fediverse posts

As of this morning: 🧠 425.49B tokens seen 📊 4.25% completed This eager reader wants more input, one token at a time. Follow along. 🔍 (2/2) #PreTraining #LLM #MultilingualAI #TransparentAI #goOpenEuroLLM

So what does this mean for multilinguality? seems like problems go deeper than we initially anticipated Got any ideas on how to fix this? Want to hear more? Feel free to contact through DMs or join our Discord server. 📬
#AIResearch #NLP #LLMs #MultilingualAI

So what does this mean for multilinguality? seems like problems go deeper than we initially anticipated Got any ideas on how to fix this? Want to hear more? Feel free to contact through DMs or join our Discord server. 📬
#AIResearch #NLP #LLMs #MultilingualAI

Deepgram released Flux Multilingual, a speech recognition model that handles 10 languages with real-time switching during conversations. The system detects language changes mid-call and processes conversational turns in under 400ms. Available as cloud API or self-hosted at the same price as English-only versions. Could simplify multilingual voice applications that previously required separate detection and routing systems.

#SpeechRecognition #MultilingualAI #VoiceTech

https://www.implicator.ai/deepgram-launches-flux-multilingual-speech-model-with-10-language-mid-call-switching/

AI’s fluency in other languages hides a Western worldview that can mislead users − a scholar of Indonesian society explains
#AI #LanguageBias #CulturalAwareness #WesternWorldview #TechEthics #Indonesia #MultilingualAI #DigitalCulture #SocialImpact #Tech
https://the-14.com/ais-fluency-in-other-languages-hides-a-western-worldview-that-can-mislead-users-%e2%88%92-a-scholar-of-indonesian-society-explains/

AI’s fluency in other languages hides a Western worldview that can mislead users − a scholar of Indonesian society explains
#AI #LanguageBias #CulturalAwareness #WesternWorldview #TechEthics #Indonesia #MultilingualAI #DigitalCulture #SocialImpact #Tech
https://the-14.com/ais-fluency-in-other-languages-hides-a-western-worldview-that-can-mislead-users-%e2%88%92-a-scholar-of-indonesian-society-explains/

👏 Congratulations on this achievement and all the best for Cecilia’s new role as postdoctoral researcher at the University of Cambridge!

#NLP #PhDDefense #MultilingualAI #CulturalAI #LanguageModels #UKPLab #TUDarmstadt Computer Science, TU Darmstadt

👏 Congratulations on this achievement and all the best for Cecilia’s new role as postdoctoral researcher at the University of Cambridge!

#NLP #PhDDefense #MultilingualAI #CulturalAI #LanguageModels #UKPLab #TUDarmstadt Computer Science, TU Darmstadt

👋 We’re happy to welcome Soumik Bhattacharjee to RC Trust! 🤖💬

Since February 1, 2026, he has joined the Human-AI Interaction group as a research assistant.

With a background in Digital Media, Soumik focuses on designing conversational AI systems that are:
🌍 multilingual
🧠 context-aware
🤝 socially intelligent

#HumanAIInteraction #ConversationalAI #AICommunity #HCI #MultilingualAI

#ArtificialIntelligence #PromptEngineering #MultilingualAI
#Localization #AIAlignment #LanguageTechnology #EnterpriseAI

#ArtificialIntelligence #PromptEngineering #MultilingualAI
#Localization #AIAlignment #LanguageTechnology #EnterpriseAI

Claude Opus 4.5 tops SWE‑bench in 7 of 8 languages, beating Sonnet 4.5 by 15%. The win spans Java, Python and other major stacks, showing Anthropic’s multilingual edge in AI‑assisted coding. Curious how the models compare? Dive into the benchmarks and see what this means for open‑source developers. #ClaudeOpus45 #Sonnet45 #SWEbench #MultilingualAI

🔗 https://aidailypost.com/news/claude-opus-45-leads-swebench-7-8-languages-15-ahead-sonnet-45

Google’s Nano Banana Pro generates 4K images with advanced editing, multilingual support, and precise controls, enhancing AI image creation for creators and businesses.

#Google #NanoBananaPro #AI #ImageGeneration #4K #ContentCreation #MultilingualAI #AIDesign #CreativeTools #TechInnovation #TECHi

Read Full Article Here :- https://www.techi.com/google-launches-nano-banana-pro/

A data-driven look at why multilingual LLMs fail at true cross-lingual reasoning, and how culturally intact datasets and new metrics reveal the gap. https://hackernoon.com/multilingual-isnt-cross-lingual-inside-my-benchmark-of-11-llms-on-mid-and-low-resource-languages #multilingualai

A data-driven look at why multilingual LLMs fail at true cross-lingual reasoning, and how culturally intact datasets and new metrics reveal the gap. https://hackernoon.com/multilingual-isnt-cross-lingual-inside-my-benchmark-of-11-llms-on-mid-and-low-resource-languages #multilingualai

Detecting Indian languages is harder than it seems—code‑mixing and scarce data trip up models. A new paper releases the first open‑source baseline ML models for language ID, fine‑tuned transformers, and a benchmark dataset covering low‑resource Indian tongues. Check out the results and tools for multilingual AI research! #LanguageID #MultilingualAI #LowResourceNLP #OpenSourceML

🔗 https://aidailypost.com/news/indian-language-id-proves-tough-authors-release-baseline-ml-models

New benchmark IndQA tackles India’s billion‑plus non‑English users, positioning the country as the world’s 2nd‑largest ChatGPT market. See how the MMMLU‑based results reshape multilingual AI research. #IndQA #ChatGPT #MultilingualAI #MMMLU

🔗 https://aidailypost.com/news/indqa-targets-indias-billion-nonenglish-users-2ndlargest-chatgpt

Polish emerges as top language in multilingual AI benchmark testing: Researchers test 26 languages across seven synthetic tasks at context lengths from 8K to 128K tokens, finding performance gaps widen at longer contexts for low-resource languages. https://ppc.land/polish-emerges-as-top-language-in-multilingual-ai-benchmark-testing/ #AI #MachineLearning #PolishLanguage #MultilingualAI #BenchmarkTesting

We Need Arabic Language Models

https://www.natureasia.com/en/nmiddleeast/article/10.1038/nmiddleeast.2025.142

#HackerNews #ArabicLanguageModels #AIResearch #LanguageTechnology #NaturalLanguageProcessing #MultilingualAI

We Need Arabic Language Models

https://www.natureasia.com/en/nmiddleeast/article/10.1038/nmiddleeast.2025.142

#HackerNews #ArabicLanguageModels #AIResearch #LanguageTechnology #NaturalLanguageProcessing #MultilingualAI

TildeOpen LLM is an open-source multilingual foundation model developed for 34 European languages. Supported by the EU and running on EuroHPC supercomputers, it aims to strengthen Europe’s digital sovereignty by providing secure, transparent and adaptable AI.

https://tilde.ai/tildeopen-llm/
#DigitalSovereignty #MultilingualAI #OpenSource

TildeOpen LLM is an open-source multilingual foundation model developed for 34 European languages. Supported by the EU and running on EuroHPC supercomputers, it aims to strengthen Europe’s digital sovereignty by providing secure, transparent and adaptable AI.

https://tilde.ai/tildeopen-llm/
#DigitalSovereignty #MultilingualAI #OpenSource

🔹 How to Auto-Detect Language In Copilot Studio
Copilot Studio now auto-detects user language so agents respond naturally and instantly. 🚀

💡 Seamless multilingual responses without manual settings.
🔍 Built-in NLU+ lets you train and test language models directly in the platform.
⚖️ One agent adapts to many languages, simplifying development and improving voice/chat UX.

▶︎ https://www.hubsite365.com/en-ww/citizen-developer/?id=ce3f6ad4-d488-f011-b4cc-7c1e52134a37&topic=9f678e9a-8cd4-ec11-a7b5-6045bd92fe52&theater=true

#CopilotStudio #NLU #MultilingualAI #AI

🔹 How to Auto-Detect Language In Copilot Studio
Copilot Studio now auto-detects user language so agents respond naturally and instantly. 🚀

💡 Seamless multilingual responses without manual settings.
🔍 Built-in NLU+ lets you train and test language models directly in the platform.
⚖️ One agent adapts to many languages, simplifying development and improving voice/chat UX.

▶︎ https://www.hubsite365.com/en-ww/citizen-developer/?id=ce3f6ad4-d488-f011-b4cc-7c1e52134a37&topic=9f678e9a-8cd4-ec11-a7b5-6045bd92fe52&theater=true

#CopilotStudio #NLU #MultilingualAI #AI

🔵 Multilanguage in Copilot Studio - Step by Step

Want to build a multilingual agent with Copilot Studio? This video walks through configuration options, typical uses, and practical trade-offs to watch for.

💡 Choose between per-language agents or a unified model
🔍 Use the GitHub samples to test patterns and edge cases.
⚖️ Plan for translation latency, prompt tuning, and localization quality.

▶︎https://www.hubsite365.com/en-ww/citizen-developer/?id=cc2c3111-807f-f011-b4cc-7c1e52134a37&topic=9f678e9a-8cd4-ec11-a7b5-6045bd92fe52&theater=true

Watch the walkthrough and explore the samples — which language are you planning to support?
#CopilotStudio #MultilingualAI #AI #DeveloperTools

🔵 Multilanguage in Copilot Studio - Step by Step

Want to build a multilingual agent with Copilot Studio? This video walks through configuration options, typical uses, and practical trade-offs to watch for.

💡 Choose between per-language agents or a unified model
🔍 Use the GitHub samples to test patterns and edge cases.
⚖️ Plan for translation latency, prompt tuning, and localization quality.

▶︎https://www.hubsite365.com/en-ww/citizen-developer/?id=cc2c3111-807f-f011-b4cc-7c1e52134a37&topic=9f678e9a-8cd4-ec11-a7b5-6045bd92fe52&theater=true

Watch the walkthrough and explore the samples — which language are you planning to support?
#CopilotStudio #MultilingualAI #AI #DeveloperTools

Learn Mandarin in your language 🌍🧠
📚 Aprender mandarín en tu idioma
📖 Apprenez le mandarin dans votre langue
🗣️ Lerne Chinesisch in deiner Sprache

Try MandarinMap GPT now:
🔗 https://tinyurl.com/4amh74dk

🧭 Just type: “Teach me in Spanish / French / German…”

#LearnMandarin #MultilingualAI #MappingMandarin #LanguageLearning #chatGPT

🇨🇭 Switzerland is taking a public-interest-first approach to LLMs. EPFL, ETH Zurich, and CSCS are building a fully open, multilingual language model trained on public infrastructure, and they’re releasing it under Apache 2.0 this summer. This one isn’t just open weights. It’s 100% transparent: source, data, training methods. And trained on 15T tokens across 1,500+ languages using a carbon-neutral supercomputer (Alps), it’s a real shot at sovereign AI that serves scientific, civic, and commercial needs without the lock-in.

TL;DR
🌍 Fluency in 1,000+ languages
🧠 Open 8B and 70B param models
⚡ Alps supercomputer, 100% green
🔓 Fully open: data, code, weights

https://ethz.ch/en/news-and-events/eth-news/news/2025/07/a-language-model-built-for-the-public-good.html
#opensourceAI #multilingualAI #sovereignAI #SwissTech #Freedom #AI

🇨🇭 Switzerland is taking a public-interest-first approach to LLMs. EPFL, ETH Zurich, and CSCS are building a fully open, multilingual language model trained on public infrastructure, and they’re releasing it under Apache 2.0 this summer. This one isn’t just open weights. It’s 100% transparent: source, data, training methods. And trained on 15T tokens across 1,500+ languages using a carbon-neutral supercomputer (Alps), it’s a real shot at sovereign AI that serves scientific, civic, and commercial needs without the lock-in.

TL;DR
🌍 Fluency in 1,000+ languages
🧠 Open 8B and 70B param models
⚡ Alps supercomputer, 100% green
🔓 Fully open: data, code, weights

https://ethz.ch/en/news-and-events/eth-news/news/2025/07/a-language-model-built-for-the-public-good.html
#opensourceAI #multilingualAI #sovereignAI #SwissTech #Freedom #AI

French AI startup Mistral AI has introduced "Magistral," a new reasoning model designed to deliver logic-based answers across multiple languages. It offers responses up to 10 times faster than competitors and provides domain-specific expertise with high accuracy for solving complex problems.

#MistralAI #Magistral #AIReasoning #MultilingualAI #AIInnovation #FutureOfAI #TechNews #ArtificialIntelligence #Greaternoida #students

🎉🚀 Behold, the latest buzzword salad 🥗 from Mistral AI, claiming to have birthed the "first" reasoning model, Magistral. Apparently, AI can now think like humans—multilingually and transparently, no less! 🤖 Just wait until you hear it try to explain why your printer won't connect. 🖨️✨
https://mistral.ai/news/magistral #MistralAI #ReasoningModel #AIInnovation #MultilingualAI #TechBuzz #HackerNews #ngated

🎉🚀 Behold, the latest buzzword salad 🥗 from Mistral AI, claiming to have birthed the "first" reasoning model, Magistral. Apparently, AI can now think like humans—multilingually and transparently, no less! 🤖 Just wait until you hear it try to explain why your printer won't connect. 🖨️✨
https://mistral.ai/news/magistral #MistralAI #ReasoningModel #AIInnovation #MultilingualAI #TechBuzz #HackerNews #ngated

UAE Launches Advanced Falcon AI Models to Lead Global Race

The UAE steps up in the global AI arena with Falcon Arabic, its first native Arabic language AI, and Falcon H1, a compact, multilingual model outperforming rivals like Meta’s LLaMA.

#UAEAI #FalconArabic #FalconH1 #ArabicAI #MultilingualAI #OpenSourceAI #TII #ATRC #AIInnovation #TechLeadership #TECHi

Read Full Article Here :- https://www.techi.com/uae-launches-falcon-arabic-ai-models-global-leadership/

Kokoro TTS – Free Text to Speech Converter with Multilingual AI Voices

https://kokoroai.org

#HackerNews #KokoroTTS #FreeSpeech #MultilingualAI #TextToSpeech #VoiceTech

Kokoro TTS – Free Text to Speech Converter with Multilingual AI Voices

https://kokoroai.org

#HackerNews #KokoroTTS #FreeSpeech #MultilingualAI #TextToSpeech #VoiceTech

AI chatbots are evolving. They now understand context and speak multiple languages, making conversations smoother and more natural. Discover how these advancements are reshaping customer interactions and what it means for your business.

Read More: https://spaculus.com/blog/ai-chatbots-2-0-multilingual-contextual-interactions/

#AI #Chatbots #MultilingualAI #CustomerExperience #Spaculus

🇯🇵 JAPAN | 🇪🇺 EU
🔴 Japan Asks EU to Build Non-English/Chinese AI

🔸 Tokyo seeks EU cooperation to reduce language bias in LLMs.
🔸 Japan sees risk in closed-source models favoring dominant languages.
🔸 METI & MIC preparing AI guidelines for businesses.
🔸 Emphasis on cross-border regulatory interoperability.

#AI #Japan #EU #LanguageBias #LLMs #ChatGPT #AIRegulation #MultilingualAI

🇯🇵 JAPAN | 🇪🇺 EU
🔴 Japan Asks EU to Build Non-English/Chinese AI

🔸 Tokyo seeks EU cooperation to reduce language bias in LLMs.
🔸 Japan sees risk in closed-source models favoring dominant languages.
🔸 METI & MIC preparing AI guidelines for businesses.
🔸 Emphasis on cross-border regulatory interoperability.

#AI #Japan #EU #LanguageBias #LLMs #ChatGPT #AIRegulation #MultilingualAI

🌍 #CohereForAI introduces #Aya, a groundbreaking #MultilingualAI research initiative

🤝 3,000+ independent researchers from 119 countries collaborating on advancing language accessibility in #AI

📚 Key achievements:
• 513M dataset covering 101 languages
• 204K original human annotations
• Three #opensource models: Aya-101, Aya Expanse 8B & 32B
• Supporting previously underserved languages

🔬 Research highlights:
• State-of-the-art performance in multilingual benchmarks
• Complete dataset and model weights openly available on #HuggingFace
• Focus on natural language understanding, summarization, and translation

⚡️ Project scope:
• 250+ language ambassadors
• 81K Discord messages
• Comprehensive research papers and technical documentation
• Named after "fern" in Twi language, symbolizing endurance

https://cohere.com/research/aya

🌍 #CohereForAI introduces #Aya, a groundbreaking #MultilingualAI research initiative

🤝 3,000+ independent researchers from 119 countries collaborating on advancing language accessibility in #AI

📚 Key achievements:
• 513M dataset covering 101 languages
• 204K original human annotations
• Three #opensource models: Aya-101, Aya Expanse 8B & 32B
• Supporting previously underserved languages

🔬 Research highlights:
• State-of-the-art performance in multilingual benchmarks
• Complete dataset and model weights openly available on #HuggingFace
• Focus on natural language understanding, summarization, and translation

⚡️ Project scope:
• 250+ language ambassadors
• 81K Discord messages
• Comprehensive research papers and technical documentation
• Named after "fern" in Twi language, symbolizing endurance

https://cohere.com/research/aya

#TechNews: #Qwen Releases New #VisionLanguage #LLM Qwen2-VL 🖥️👁️

After a year of development, #Qwen has released Qwen2-VL, its latest #AI system for interpreting visual and textual information. 🚀

Key Features of Qwen2-VL:

1. 🖼️ Image Understanding:

Qwen2-VL shows performance on #VisualUnderstanding benchmarks including #MathVista, #DocVQA, #RealWorldQA, and #MTVQA.

2. 🎬 Video Analysis:

Qwen2-VL can analyze videos over 20 minutes in length. This is achieved through online streaming capabilities, allowing for video-based #QuestionAnswering, #Dialog, and #ContentCreation. #VideoAnalysis

3. 🤖 Device Integration:

The #AI can be integrated with #mobile phones, #robots, and other devices. It uses reasoning and decision-making abilities to interpret visual environments and text instructions for device control. #AIAssistants 📱

4. 🌍 Multilingual Capabilities:

Qwen2-VL understands text in images across multiple languages. It supports most European languages, Japanese, Korean, Arabic, Vietnamese, among others, in addition to English and Chinese. #MultilingualAI

This release represents an advancement in #ArtificialIntelligence, combining visual perception and language understanding. 🧠 Potential applications include #education, #healthcare, #robotics, and #contentmoderation.

https://github.com/QwenLM/Qwen2-VL

#TechNews: #Qwen Releases New #VisionLanguage #LLM Qwen2-VL 🖥️👁️

After a year of development, #Qwen has released Qwen2-VL, its latest #AI system for interpreting visual and textual information. 🚀

Key Features of Qwen2-VL:

1. 🖼️ Image Understanding:

Qwen2-VL shows performance on #VisualUnderstanding benchmarks including #MathVista, #DocVQA, #RealWorldQA, and #MTVQA.

2. 🎬 Video Analysis:

Qwen2-VL can analyze videos over 20 minutes in length. This is achieved through online streaming capabilities, allowing for video-based #QuestionAnswering, #Dialog, and #ContentCreation. #VideoAnalysis

3. 🤖 Device Integration:

The #AI can be integrated with #mobile phones, #robots, and other devices. It uses reasoning and decision-making abilities to interpret visual environments and text instructions for device control. #AIAssistants 📱

4. 🌍 Multilingual Capabilities:

Qwen2-VL understands text in images across multiple languages. It supports most European languages, Japanese, Korean, Arabic, Vietnamese, among others, in addition to English and Chinese. #MultilingualAI

This release represents an advancement in #ArtificialIntelligence, combining visual perception and language understanding. 🧠 Potential applications include #education, #healthcare, #robotics, and #contentmoderation.

https://github.com/QwenLM/Qwen2-VL

model paper https://cohere.com/research/papers/aya-model-paper-2024-02-13
dataset paper https://cohere.com/research/papers/aya-dataset-paper-2024-02-13
dataset on hF https://huggingface.co/datasets/CohereForAI/aya_collection
model on HF https://huggingface.co/CohereForAI/aya-101
#AI #MultilingualAI #OpenScience #CohereForAI #MachineLearning