Qwen team
PulseAugur coverage of Qwen team — every cluster mentioning Qwen team across labs, papers, and developer communities, ranked by signal.
1 day(s) with sentiment data
-
Qwen releases Qwen-Image-2.1 for unified image generation and editing
The Qwen team has released Qwen-Image-2.1, an open-source model that integrates text-to-image generation and editing capabilities. This new version features a more compact 7B-parameter DiT architecture for image generat…
-
GLM-5.3 and Qwen models see new releases and local execution methods · 5 sources tracked
The GLM-5.3 model is set to be released, with its earlier version, Ox Alpha, being an initial iteration that offered less performance and stability compared to the official release. Concurrently, the Qwen team has devel…
-
New AI honeypot 'Chameleon' uses LLMs to adapt to threats
Researchers have developed Chameleon, an adaptive AI-driven honeypot architecture designed to overcome the limitations of traditional honeypots. This new platform integrates a BiLSTM classifier for threat detection, a Q…
-
Together AI launches Qwen3.8-2.4T-A95B for coding and agentic tasks
Together AI has launched Qwen3.8-2.4T-A95B, a new flagship model from the Qwen team, on its Serverless Inference platform. This model is designed for coding and long-horizon agent workflows, boasting 2.4 trillion parame…
-
Uncensored MiniMax-H3 text encoder optimized for 16GB GPUs released
A new, uncensored text encoder for the MiniMax-H3 video generation model has been released, optimized to fit on a single 16GB GPU. This quantized version, named Qwen3-VL-32B Heretic (MiniMax-H3 text encoder) — NVFP4, is…
-
Quantized Qwen3-VL-32B-Heretic models released for MiniMax-H3 and H3 Healthcare
New quantized versions of the Qwen3-VL-32B-Heretic model are now available, specifically tailored for the MiniMax-H3 and H3 Healthcare Three Hop Index text encoders. These versions, including an NVFP4 quantization and a…
-
Qwen3 LLM runs up to 4.52x faster on Apple Silicon with ExecuTorch MLX delegate
A recent technical exploration demonstrates significant speed improvements when running the Qwen3-0.6B language model on Apple Silicon using ExecuTorch's experimental MLX delegate. The MLX delegate, which leverages Appl…
-
Alibaba's Qwen3-Coder-Next achieves 70.6% on SWE-bench with efficient MoE architecture
The Qwen3-Coder-Next model, an 80 billion parameter Mixture-of-Experts model from Alibaba's Qwen team, has demonstrated impressive efficiency by achieving 70.6% on the SWE-bench Verified benchmark with only approximatel…
-
New decoding strategy bypasses LLM alignment tax for better reasoning
Researchers have introduced a novel decoding strategy called Confident Decoding, which aims to mitigate the "alignment tax" in large language models. This tax occurs when final layers of LLMs, after being fine-tuned for…
-
Alibaba Qwen3.5 model offers real-time translation with voice cloning
Alibaba's Qwen team has released Qwen3.5-LiveTranslate-Flash, a real-time multimodal translation model that significantly reduces latency to 2.8 seconds. This new model expands language support to 60 input languages and…
-
CodePercept boosts LLM visual perception using code, not just reasoning
Researchers from Shanghai Jiao Tong University and the Qwen team have introduced CodePercept, a novel approach to enhance large language models' visual perception capabilities, particularly for STEM tasks. Their researc…