Tencent’s Hy4 AI Beats DeepSeek in Blind Test by 213 Finance Pros
Tencent’s new Hy4 AI model beat open-source rivals in a finance expert blind test while keeping prices flat and adding free access offers.
Stay ahead of the curve with the latest AI news, breakthroughs, product launches, research, funding, and insights from across the artificial intelligence industry.
Tencent’s new Hy4 AI model beat open-source rivals in a finance expert blind test while keeping prices flat and adding free access offers.
Mistral released Timestral, a family of four time-series forecasting models whose largest version ranks first on three major public benchmarks.
Anthropic says Alibaba, Moonshot AI, and DeepSeek ran large distillation campaigns to copy Claude's reasoning and coding abilities.
OpenAI released GPT-6 Astra, a new AI model with top scores on math, coding, and cybersecurity benchmarks, rolling out across ChatGPT and cloud platforms this week.
OpenAI launched GPT-6 Astra, a new AI model with top benchmark scores, faster computer use, and stronger safety alignment than earlier versions.
OpenAI's GPT-6 Astra excels at 3D design, coding and computer use tasks, but early testers found its writing quality fell behind its predecessor.
Google released WeatherNext 3, an AI forecasting model that beats government and rival AI weather systems on accuracy and speed.
OpenAI released its Astra AI model, calling it a step toward general intelligence, while flagging its highest-ever internal cybersecurity risk level.
OpenAI's new Astra model reportedly uses a reasoning method that could make AI thinking harder to monitor, worrying safety researchers.
Google launched Gemini 3.8 Flash and Flash Cyber, new AI models for coding, reasoning, and cybersecurity defense, at the same price as their predecessor.
Z.ai confirmed it created the mystery model Ox Alpha, with public weights set to release Wednesday amid rising competition from Chinese AI labs.
OpenAI unveiled test results for Jalapeño, its first custom AI chip, showing faster and more power-efficient performance than existing hardware.