AI BriefingQwen news
The latest AI and developer news from Qwen, with short summaries.
Visit website →Qwen-Image: Image Creation with Native Text Rendering
Qwen-Image is a 20B image model strong in complex text rendering and precise editing.
GSPO for Scalable Reinforcement Learning of Language Models
GSPO reduces GRPO's instability through sequence-level optimization, enabling RL scaling.
Qwen-MT: Translation Where Speed Meets Intelligence
Qwen-MT boosts performance with 92-language translation, customization, and low-cost, high-speed processing.
Qwen3-Coder: Agentic Coding for the World
The 480B MoE Qwen3-Coder aims for SOTA in coding, browser use, and tool use.
Qwen-TTS Now Speaks Chinese Dialects Too
Qwen-TTS now supports 3 Chinese dialects and Chinese-English bilingual speech.
Qwen VLo: From 'Understanding' to 'Expression'
Qwen VLo is a multimodal model that combines image understanding and generation into one.
Qwen3 Embedding: Advancing Text Embedding and Reranking with Foundation Models
The Qwen3-based embedding and reranker series has boosted multilingual retrieval performance.
Qwen3: Thinking Deeper and Acting Faster
Qwen3 is an open-weight model with a hybrid thinking mode and support for 119 languages.
QVQ-Max: Thinking with Evidence
The visual reasoning model **QVQ-Max**, which analyzes visual information and reasons to present solutions, has been released.
Qwen2.5 Omni: An All-in-One Multimodal Model That Sees, Hears, Speaks, and Writes
Qwen2.5-Omni, the new flagship model in the Qwen series, is an end-to-end multimodal model that unifiedly processes text, images, audio, and video.
Qwen2.5-VL-32B: A Smarter, Lighter Model
Qwen has released Qwen2.5-VL-32B-Instruct, an open-source multimodal model at 32B scale optimized for performance through reinforcement learning.
QwQ-32B: Harnessing the Power of Reinforcement Learning
Qwen released **QwQ-32B**, a model that achieves reasoning performance comparable to DeepSeek-R1 through reinforcement learning.
QwQ-Max-Preview Released
Qwen has released **QwQ-Max-Preview**, which enhances reasoning and coding performance, and has previewed future app launches and open-sourcing of small models.
Qwen2.5-Max: Exploring the Intelligence of a Large-Scale MoE Model
Alibaba unveiled **Qwen2.5-Max**, a large-scale MoE model trained on over 20 trillion tokens.
Qwen2.5-1M: Qwen Releases Models Supporting Up to 1 Million Token Context Length
Qwen has released the Qwen2.5-1M series, open-source models supporting up to 1 million tokens of context length.
Qwen2.5-VL Released
Qwen has released Qwen2.5-VL, its next-generation vision-language model with agentic capabilities and long-video understanding.
An Almost-Free Method to Improve MoE LLM Training Performance: Global-Batch Load Balancing
Applying load balancing at the global-batch level instead of the micro-batch level significantly improves MoE model performance and expert specialization.
Towards Effective Process Supervision for Mathematical Reasoning
Qwen has released a new PRM and benchmark called ProcessBench that identifies step-by-step errors in mathematical reasoning.
QVQ: Looking at the World with Wise Sight
The Qwen team has released QVQ, an open-weight multimodal model with enhanced visual reasoning capabilities.
QwQ: Deep Reflection at the Edge of the Unknown
Qwen Team has released **QwQ-32B-Preview**, an experimental model with enhanced deep reasoning capabilities in mathematics and coding.