Flash

The news, the context, and what changes.

RSS
Flash

StepFun Launches StepAudio 3 Series of Voice Large Models

StepFun has released the StepAudio 3 series of voice large models, covering five areas: real-time interaction, speech recognition, speech synthesis, full audio generation, and music creation. The Realtime model topped the global real-time voice interaction rankings with a comprehensive score of 98.9%, while the ASR model achieved a word error rate of just 1.7%, tying for first place. Several models in the series have achieved leading global positions on the Artificial Analysis leaderboard, demon

88Z AI editorial1 min read
Flash

HiDream Future Launches Omni-Modal Video Generation Model HiDream-O1-Video-1.0

HiDream Future has released its first native omni-modal video model, HiDream-O1-Video-1.0, which supports multi-modal inputs and generates 5-20 second 1080p videos with one click. The model features significant upgrades in intent understanding, physics simulation, and audio-visual synchronization, ranking highly on global authoritative lists and marking the entry of Chinese AI video technology into the world's top tier.

88Z AI editorial1 min read
Flash

National Supercomputing Internet Platform Has Launched Over 400 Agents

He Mujun, President of Sugon Intelligent Computing, revealed at the 2026 Computing Power Network Development Conference that the Supercomputing Internet platform has over 1.7 million registered users. It aggregates 3.5 million CPU cores and 250,000 GPU cards, with domestic computing power accounting for more than 99%. The platform has built six major agent matrices, launched over 400 agents and 10,000 skills, and partnered with dozens of institutions including Peking University and BGI.

88Z AI editorial1 min read
Flash

Moonshot AI Launches New Flagship Model K2.8 Preview

Moonshot AI has released its new flagship model, K2.8 Preview, fully integrated into Kimi Code and Kimi Work. Supporting a 1M context window, the model features enhanced coding and agent capabilities, with overall performance approaching the flagship Kimi K3. It is available to all membership tiers. Previously, Kimi K3's state-of-the-art frontend capabilities helped drive the company's ARR from $300 million in June to over $1 billion in August.

88Z AI editorial1 min read
Flash

Zhipu Announces Completion of Approximately $5 Billion Financing

Zhipu announced the completion of approximately $5 billion in financing, comprising a $2 billion equity placement and $3 billion in zero-coupon convertible bonds. The funds will prioritize R&D for the next-generation GLM foundation models and computing infrastructure, with 60% allocated to self-training systems and recursive model improvement. In the first half of the year, revenue reached 954 million yuan, up nearly 400% year-on-year, with an ARR of $1.6 billion.

88Z AI editorial1 min read
Flash

Ant Lingbo Technology Open Sources Real-time Interactive World Model LingBot-World 2.0

Following the open-source release of its 14B main model, Ant Lingbo Technology has launched three additional models for the LingBot-World 2.0 series. These include a Small version for consumer GPUs, a bidirectional Teacher model for high-quality distillation, and a causal pre-training base supporting multi-modal inputs, aiming to lower deployment barriers and improve training efficiency.

88Z AI editorial1 min read
Flash

SenseTime's Xiao Huanxiong Launches on Mobile

SenseTime's Xiao Huanxiong mobile app has officially launched, supporting both iOS and Android platforms. The application features core capabilities such as controlling computers via mobile commands, remotely viewing and sharing results, and long-press voice input, enabling seamless cross-device collaboration.

88Z AI editorial1 min read
Flash

ChatGPT Voice Mode Supports GPT-5.6 Sol and GPT-6 Astra

OpenAI has updated ChatGPT Voice Mode, allowing Pro users to select between GPT-5.6 Sol and GPT-6 Astra models and adjust reasoning intensity during voice conversations. The system automatically invokes the chosen model for search or complex reasoning tasks, with each invocation counting toward message limits.

88Z AI editorial1 min read
Flash

Kimi Officially Launches Corporate Partner Program

Moonshot AI's Kimi has launched a corporate partner program, collaborating with IT service providers and integrators to form frontline deployment teams that drive AI into core business operations. Companies including Huasheng Tiancheng and Kingsoft Cloud have signed on to co-build industry solutions.

88Z AI editorial1 min read
Flash

DeepSeek Officially Open Sources DeepSeek V4.1 Flash

DeepSeek has officially open-sourced DeepSeek V4.1 Flash, a Mixture of Experts model with 552 billion parameters. It features an asymmetric causal encoder-decoder architecture, with only 8 billion input activation parameters and 16 billion for output, along with native multimodal capabilities. By significantly compressing the KV cache, it reduces costs for agent scenarios. The API is now live, and Tencent's WorkBuddy, CodeBuddy, and OpenCode have fully integrated the model.

88Z AI editorial1 min read