Flash

The news, the context, and what changes.

RSS
Flash

Zhipu Launches High-Speed Inference Model GLM-5.3-FlashX

Zhipu has released GLM-5.3-FlashX, achieving inference speeds of up to 200 tokens/s. Previously introduced as Ox Alpha, the model is built on the computing power of 100,000 domestic chips, with enhanced infrastructure investment and inference optimization. As a leading model in intelligence for its size, the GLM-5.3-Flash series offers high cost-effectiveness, and this upgrade further strengthens its comprehensive competitiveness in intelligence, price, and speed.

88Z AI editorial1 min read
Flash

Tencent Officially Launches AI Game Creation Platform 'Youdianling'

Tencent's AI game creation tool, formerly codenamed 'Craft', is officially named 'Youdianling CraftBuddy'. The new name retains 'Craft' and incorporates the concept of a partner, positioning itself as the user's 'AI game creation companion'. Users can visit craft.qq.com to experience games created with Youdianling. Creation access is being rolled out gradually, and official channels should be followed for event updates.

88Z AI editorial1 min read
Flash

Tencent Open-Sources Self-Hosted AI Assistant Octop

Tencent Cloud has released Octop 1.0 GA, a self-hosted multi-agent AI assistant designed for solopreneurs, families, and small teams. The update enhances knowledge retention, multi-device access, permission controls, and security. It is now available on the Lighthouse and CVM official image markets, enabling one-command deployment of a private AI collaboration hub.

88Z AI editorial1 min read
Flash

Tencent Launches AI-Native Voice Assistant Chatterfly

Tencent has launched Chatterfly, an AI-native voice assistant, marking its entry into the AI input sector. Positioned as an expression aid, the tool automates the process from spoken ideas to polished text. It is optimized for weak network and noisy environments and integrates with Agent products within the Tencent ecosystem.

88Z AI editorial1 min read
Flash

Kimi Releases Financial Industry Solution

Moonshot AI's Kimi has launched a financial industry solution that integrates multiple data sources and professional skills. In partnership with China Securities Co., Ltd., it has significantly reduced report production time. The solution is now fully online and adopted by numerous institutions.

88Z AI editorial1 min read
Flash

iFlytek Launches All-Scenario AI Productivity Workbench AStudio

iFlytek Open Platform has officially launched AStudio, a desktop AI workbench. Built on the Spark X2.5 model and equipped with the ACode Harness Agent framework, it supports multi-model switching, planning modes, and multi-agent collaboration. Featuring dual spaces for daily tasks and projects, AStudio connects to local files and real-world projects to generate, deliver, and verify documents, web pages, and code, aiming to enhance AI productivity across all scenarios.

88Z AI editorial1 min read
Flash

Google Launches Native Real-Time Voice Conversation Model Gemini 3.8 Live

Google has released the Gemini 3.8 Live series of real-time conversation models, comprising a standard version and an Extended Thinking version. The standard edition focuses on scalable deployment and cost optimization, supporting automatic multi-language switching and visual understanding. The Extended Thinking edition enhances multi-step reasoning capabilities, enabling simultaneous reasoning and speech, and performs excellently in various benchmark tests.

88Z AI editorial1 min read
Flash

Doubao Work Model Update and Launch of 'Doubao Work Partner'

Doubao Work announces two major updates: the Doubao 2.1 Pro model has completed a minor version upgrade and is now fully available via Volcano Ark API, while the new 'Doubao Work Partner' independently integrates into Feishu for on-demand team collaboration.

88Z AI editorial1 min read
Flash

Shanghai AI Lab Open-Sources Multimodal Foundation Model 'ShuSheng-S2'

Shanghai AI Laboratory has released ShuSheng-S2, an open-source multimodal foundation model. Its general capabilities rank among the top tier of open-source models. The model performs comparably to leading closed-source systems in scientific long-horizon tasks such as biology, materials science, chemistry, and advanced mathematics. By employing a Memory Decoder architecture with pluggable professional memory modules, ShuSheng-S2 balances domain expansion with general proficiency while strengthen

88Z AI editorial1 min read
Flash

Unitree Open-Sources General Humanoid Foundation Model UnifoLM-WLA-1.0

Unitree has fully open-sourced its new general humanoid foundation model, UnifoLM-WLA-1.0, releasing the code, model weights, and datasets simultaneously. Trained on large-scale multimodal perception and interaction-centric world modeling using approximately 2,500 hours of real-robot data, the model achieves leading performance in embodied reasoning benchmarks, rivaling top closed-source models. It enables a single model to manage 64 tasks, covering both desktop manipulation and whole-body mobil

88Z AI editorial1 min read