Microsoft Fara1.5 open source computer operation Agent: 27B small model beats OpenAI Operator and Gemini 2.5
The Microsoft Research AI Frontiers team released a completely open source computer operation A
Tag: HuggingFace
Microsoft open source Mage-Flow image model: only 4B parameters can generate 1024×1024 images in one second, and the benchmark test is tied with FLUX.2 32B
Microsoft Asia publicly released Mage-Flow on July 22 at Hugging Face
Baidu launches open-source OCR model Unlimited-OCR with 3B parameters, outperforming GPT-5 level models
Chinese AI company Baidu recently open-sourced a new tool called "Unlimited-OCR" on Hugging Face
NVIDIA releases Nemotron 3.5 ASR, a compact 0.6B speech recognition model that can transcribe 40 languages in real-time on CPU only
NVIDIA's NeMo speech team quietly released a 600M parameter automatic speech recognition model called Nemo Tr...
Microsoft Open-Sources VibeVoice Speech AI Model: Transcribe 60-Minute Long Audio in One Take, 90-Minute Multi-Speaker TTS Transcription
Microsoft Research Asia recently released the open-source VibeVoice series of speech AI models. This suite of models covers Automatic Speech ...
Xiaomi releases the MiMo-V2.5 series of large models, and the open-source version has claimed the top spot on the global open-source leaderboard.
Xiaomi only unveiled the Mimo-V2 large model last month, and not long after, it recently announced the updated and more ...
Baidu Open-Sources 8B Text-to-Image Model ERNIE-Image: Runs on 24GB GPUs, Leads Peers in Text Rendering and Poster Layout
After Alibaba, Tencent, MiniMax and several other major companies successively released diffusion models that can run on consumer-grade hardware, China's ...
MediaTek Innovation Center Launches Breeze 3 Series: Taiwan’s First Open-Source Taiwanese Hokkien Speech Recognition Model Breeze-ASR-26, Enabling AI to Understand Taiwanese
Previous ASR (speech-to-text) systems could generally only recognize Mandarin. If you said: "Your kha-bang is so beautiful" (the bag is ...
VoxCPM2 Open-source Voice Model: Voice Cloning Tool Free from Tokenizer Constraints, Runs with Just 8GB VRAM
For a long time, speech synthesis technology has largely relied on discretizing sound into "Tokens" - a fragmented approach that, ...
MiniMax Officially Open-Sources M2.7: Autonomous Training for 100 Rounds, Fixes Production Failures in Under 30 Minutes
Chinese AI startup MiniMax (Xiyu Technology), which has received high praise in the open-source community, officially open-sourced its latest on ...