AI News
Daily curated AI news: breakthroughs, product launches, industry trends and more.
Latest News
Total 157 articles
AI News (2026/5/6): OpenAI Launches New Default Model GPT-5.5 Instant for ChatGPT
OpenAI has officially launched GPT-5.5 Instant, replacing GPT-5.3 Instant as the default model for ChatGPT. The new model reduces the hallucination rate by 52.5% in professional fields such as healthcare, law, and finance, and significantly improves scores in math competitions. Additionally, responses are more concise and natural, reducing unnecessary formatting. Plus/Pro users can call up historical conversations and files for personalized responses, and all consumer versions will add a new "memory source" feature for users to view and manage.
AI News (2026/5/4): xAI Launches Latest Flagship Inference Model Grok 4.3
xAI has launched the Grok 4.3 model, positioned as a pragmatic transitional version. The model's API prices have been reduced by 40%-60%, with an output speed of 196 Tokens/s and support for 1 million Token long context. It has shown significant improvements in agent tasks and office assistance, particularly in generating documents, tables, and presentations. However, the model lags behind GPT-5.5 and Claude Opus 4.7 in the Intelligence Index, with insufficient stability in complex reasoning and fact-checking, and an increased hallucination rate.
AI News (2026/4/30): Alibaba Launches Two Agent Products: QoderWake and Qoder Mobile
Alibaba Cloud has officially launched the QoderWake digital employee platform and the Qoder mobile app. QoderWake uses a Harness-First architecture, which allows for multi-dimensional self-evolution, enabling it to retain experience across five dimensions: memory, skills, knowledge, behavior, and feedback. This addresses the issue of general-purpose Agents forgetting tasks once they are completed. Currently, QoderWake has introduced a "Digital Programmer" feature that automates root cause analysis and code repair within Alibaba, reducing the time spent on each issue from 30 minutes to 2 minutes.
AI News (2026/4/30): Tencent Hunyuan Open-Sources Offline Translation Model for Mobile...
Tencent Hunyuan has open-sourced the offline translation model Hy-MT1.5-1.8B-1.25bit for mobile devices, supporting 33 languages and 1056 translation directions. The model is compressed to 440MB using Sherry sparse quantization technology and can run locally on a mobile device without an internet connection. The translation quality surpasses mainstream systems like Google Translate.
AI News (2026/4/29): SenseNova U1 Series of Native Understanding and Generation Unified...
SenseTime has officially open-sourced the SenseNova U1 series of native understanding and generation unified models. These models are based on the self-developed NEO-unify architecture, which eliminates traditional concatenation design and achieves unified multi-modal understanding, reasoning, and generation within a single framework. The lightweight version, SenseNova U1 Lite, includes two specifications: an 8B dense network and an A3B MoE network, both of which achieve state-of-the-art (SOTA) performance in various benchmark tests.
AI News (2026/4/28): Microsoft GitHub Copilot Announces Pay-As-You-Go Billing
GitHub Copilot announced that starting from June 1, the billing model will change from fixed quotas to pay-as-you-go, introducing AI Credits as the new billing unit. The base subscription prices remain unchanged: Pro at $10 per month and Pro+ at $39 per month, both including equivalent AI Credits. The code completion feature will not consume any credits. Enterprise editions support shared credit pools, and Business and Enterprise customers will receive additional promotional credits from June to August.
AI News (2026/4/27): LibTV Launches HappyHorse 1.0 Video Model
LibTV has launched the new HappyHorse 1.0 video model, which supports three modes: text-to-video, image-to-video, and reference image generation. It can output 15-second 1080P multi-shot narrative videos. The model features natural language video editing, cinematic-quality visuals, intelligent shot arrangement, synchronized audio and video generation, and the ability to reproduce multiple styles, significantly improving the naturalness of character movements, micro-expressions, and dialogue realism.
AI News (2026/4/27): DeepSeek API Input Cache Price Reduced by 90%
DeepSeek officially announced that starting from April 27, the price for input cache hits across all API services will be reduced to one-tenth of the original price. The cache hit input price for DeepSeek-V4-Pro will be reduced to 0.025 yuan, and for DeepSeek-V4-Flash to 0.02 yuan. Pro models will also enjoy a 2.5-fold discount until May 5. This move aims to reduce the cost for developers and improve the cost-effectiveness of API usage in long-context scenarios.
AI News (2026/4/24): DeepSeek Launches DeepSeek-V4 Preview Version
DeepSeek has officially launched the preview version of its new model series, DeepSeek-V4, and has also open-sourced it. The series includes deepseek-v4-pro and deepseek-v4-flash, both of which support 1M ultra-long context. V4-Pro matches the performance of top-tier closed-source models in agent encoding, world knowledge, and reasoning, while V4-Flash provides similar reasoning capabilities at a lower cost. The models adopt a new attention mechanism and DSA sparse attention, significantly reducing the computational and memory overhead for long contexts.
AI News (2026/4/23): OpenAI Launches Team Collaboration Agents Workspace Agents
OpenAI has announced the launch of Workspace Agents in ChatGPT, supporting teams in creating collaborative agents to handle complex tasks and long-term workflows. This feature is powered by Codex and includes capabilities such as file handling, code execution, tool invocation, and memory storage, allowing 24/7 cloud operation. The system supports Slack integration and scheduled tasks, and is currently available only to ChatGPT Business, Enterprise, Edu, and Teachers plan users.
Subscribe to Daily AI News
Get curated AI news delivered to your inbox daily. Never miss an update.
We never share your email. Unsubscribe anytime.