Tencent

Latest AI news, models and releases from Tencent. ['AI Navigator+', 'GR', 'Hunyuan', 'HUNYUAN', 'Hunyuan 3', 'Hunyuan 3D', 'Hunyuan 3 (混元3)', 'Hunyuan Hy3', 'Hunyuan T1I', 'Hunyuan T1P', 'Hunyuan T1R', 'Hunyuan-TurboS', 'HunyuanVideo', 'HunyuanVideo 1.5', 'Hy2', 'Hy3', 'Hy3 (free)', 'Hy-Embodied-0.5-VLA', 'Hy-Embodied-RxBrain-1.0', 'Hy-Embodied-VLM-1.0', 'Hyra', 'Hyra-1.0', 'Marvis', 'Miora', 'not specified', 'Robotic Composition Model', 'T1', 'Tencent Hunyuan', 'WeLM', 'WorkBuddy', 'Yuanbao', '腾讯云 (Tencent Cloud)']

Models 🇺🇸

From GPT-2 to gpt-oss: Analysis of Architectural Improvements and Comparison with Qwen3

OpenAI has released its first open-weight models since 2019 — gpt-oss-120b and gpt-oss-20b. The article examines key architectural changes compared to GPT-2: removal of dropout, replacement of absolute positional embeddings with RoPE, transition from GELU to SwiGLU, introduction of mixture of experts (MoE) and grouped query attention (GQA). It also discusses MXFP4 optimization for running on a single GPU and comparison with Qwen3 and GPT-5.

OpenAIOpenAI Alibaba/QwenAlibaba/Qwen MetaMeta Google/DeepMindGoogle/DeepMind AI21 LabsAI21 Labs TencentTencent
Sebastian Raschka27.07 · 18:03
Fresh news