AI Briefing#multilingual
The latest AI and developer news about #multilingual, with the original source and a short summary.
The latest AI and developer news about #multilingual, with the original source and a short summary.
Pocket TTS has released a multilingual model supporting 6 languages along with an ONNX export tool.
IBM has unveiled Granite-4.1-30B, a 30B-scale model with enhanced tool calling and multilingual capabilities.
The 2026 MLC-SLM Challenge has opened registration for a multilingual Speech LLM task with a $20,000 prize pool.
Turbo-OCR added a PP-StructureV3 layout model and multilingual OCR.
An open-source 82M-scale model that detects and removes personally identifiable information (PII) from multilingual text has been released.
IBM has released Granite-4.1-8B, enhancing tool calling and instruction-following performance.
8 languages and math/code converged into the same internal space in the middle layers.
They trained a fast and accurate 6-language OCR model on 12.2 million synthetic data images.
mii-llm released the 0.4B multilingual models Zagreus/Nesso, trained on 64×A100.
Added 8 Indian languages to Chatterbox-Multilingual using only LoRA and tokenizer expansion.
A 2B-scale TTS/voice cloning model supporting 30 languages and 48kHz output.
A 0.1B-scale multilingual TTS model supports real-time generation even on a 4-core CPU.
Scribe v2 Realtime delivers STT for real-time agents with ultra-low latency of 30-80ms.
Qwen3-TTS has been released with a 1.7B and 0.6B lineup, supporting voice design, voice cloning, and streaming generation.
Qwen3-TTS-Flash now supports 49 voices, 10 languages, and 9 dialects.
HuggingFace announced an update to the Open ASR Leaderboard that adds multilingual and long-form tracks, along with the latest trends in speech recognition.
Qwen3Guard is a safety guardrail that judges prompts and responses in real time.
mmBERT, a next-generation multilingual encoder based on the ModernBERT architecture and supporting over 1,800 languages, has been released.
Google has released EmbeddingGemma, a 308M-scale multilingual embedding model optimized for on-device environments.
Qwen-MT boosts performance with 92-language translation, customization, and low-cost, high-speed processing.
HuggingFace has unveiled SmolLM3, which delivers outstanding performance at the 3B scale and supports a 128k long context.
The Qwen3-based embedding and reranker series has boosted multilingual retrieval performance.
Qwen3 is an open-weight model with a hybrid thinking mode and support for 119 languages.
Google has released Gemma 3, a new open model supporting multimodality and a 128k context.