Discover quality AI tools
Bring together the world’s leading artificial intelligence tools to help improve work efficiency
Articles published by NicholasLewis
Kouming Wanwei at WAIC 2026 declared 2026 the Year of the World Model, launching Matrix-Game3.5 with patch-level memory for real-time 20FPS 720p generation on a single GPU, plus Mureka v9.5 and O3 music models enabling versioned AI music creation. The world model core architecture is open-sourced.
Kouming Wanwei at WAIC 2026 declared 2026 the Year of the World Model, launching Matrix-Game3.5 with patch-level memory for real-time 20FPS 720p generation on a single GPU, plus Mureka v9.5 and O3 music models enabling versioned AI music creation. The world model core architecture is open-sourced.
Xiaomi slashes prices for its MiMo-V2.5 large language model API by up to 99% and upgrades its billing system. The permanent cuts, effective May 27, aim to drastically lower developer costs. Xiaomi credits the reductions to breakthroughs in its inference architecture, including optimized caching and cluster throughput. This move intensifies the ongoing price war in the AI model sector.
Xiaomi slashes prices for its MiMo-V2.5 large language model API by up to 99% and upgrades its billing system. The permanent cuts, effective May 27, aim to drastically lower developer costs. Xiaomi credits the reductions to breakthroughs in its inference architecture, including optimized caching and cluster throughput. This move intensifies the ongoing price war in the AI model sector.
Chinese AI firm Zhipu AI launched its GLM-5.1 Highspeed API for select enterprises, achieving a breakthrough output speed of 400 tokens per second. This surpasses current global API speed limits, defying the industry trade-off between speed and model capability. The model, developed through deep system-level optimization, enables flagship performance with low latency for production use in scenarios like AI programming and real-time interaction.
Chinese AI firm Zhipu AI launched its GLM-5.1 Highspeed API for select enterprises, achieving a breakthrough output speed of 400 tokens per second. This surpasses current global API speed limits, defying the industry trade-off between speed and model capability. The model, developed through deep system-level optimization, enables flagship performance with low latency for production use in scenarios like AI programming and real-time interaction.
Anthropic introduces Dreaming feature for Claude AI, simulating REM sleep to optimize memory by merging duplicates and updating outdated data from up to 100 past conversations. The system improves task success rates, notably boosting legal tech performance sixfold, and is supported by a new SpaceX supercomputing center with 220,000 GPUs.
Anthropic introduces Dreaming feature for Claude AI, simulating REM sleep to optimize memory by merging duplicates and updating outdated data from up to 100 past conversations. The system improves task success rates, notably boosting legal tech performance sixfold, and is supported by a new SpaceX supercomputing center with 220,000 GPUs.
AI adoption is surging in the gaming industry, with an estimated 90% of developers using AI tools, often discreetly. Google Cloud's gaming head notes companies like Ubisoft and Capcom leverage generative AI (e.g., Gemini) to accelerate development, handle minor assets, and reduce costs, enabling faster releases and greater creative risk-taking.
AI adoption is surging in the gaming industry, with an estimated 90% of developers using AI tools, often discreetly. Google Cloud's gaming head notes companies like Ubisoft and Capcom leverage generative AI (e.g., Gemini) to accelerate development, handle minor assets, and reduce costs, enabling faster releases and greater creative risk-taking.
AI chip firm Cerebras secures a major three-year deal exceeding $2 billion with OpenAI, which also invests roughly $1 billion for data center development and gains warrants for up to a 10% stake. Cerebras is preparing for an IPO targeting over $35 billion valuation to raise about $3 billion, strengthening its position in the competitive AI hardware market.
AI chip firm Cerebras secures a major three-year deal exceeding $2 billion with OpenAI, which also invests roughly $1 billion for data center development and gains warrants for up to a 10% stake. Cerebras is preparing for an IPO targeting over $35 billion valuation to raise about $3 billion, strengthening its position in the competitive AI hardware market.
Tencent's Hunyuan team open-sourced DisCa, a novel acceleration solution for AI video generation. It uses a learnable feature caching technique to speed up inference on distilled models by 11.8x without quality loss. The work, accepted by CVPR 2026, also includes improvements to the MeanFlow method for more stable video model training, now used in HunyuanVideo-1.5.
Tencent's Hunyuan team open-sourced DisCa, a novel acceleration solution for AI video generation. It uses a learnable feature caching technique to speed up inference on distilled models by 11.8x without quality loss. The work, accepted by CVPR 2026, also includes improvements to the MeanFlow method for more stable video model training, now used in HunyuanVideo-1.5.
Panasonic restructures leadership to boost AI integration and solution profitability. It appoints its first Chief AI Officer, Akira Sakakibara, to drive AI adoption for customer solutions and internal efficiency. Hirofumi Suzuki becomes Solution Revenue Officer to streamline cross-domain initiatives and enhance profit in key solution areas. The changes are effective April 1.
Panasonic restructures leadership to boost AI integration and solution profitability. It appoints its first Chief AI Officer, Akira Sakakibara, to drive AI adoption for customer solutions and internal efficiency. Hirofumi Suzuki becomes Solution Revenue Officer to streamline cross-domain initiatives and enhance profit in key solution areas. The changes are effective April 1.
Adobe adds new AI features to Acrobat and its collaborative Spaces tool, enabling users to generate podcast summaries, create presentations from document content, and edit files using text prompts. This expands its suite of AI capabilities amid growing competition from tools like Canva and NotebookLM.
Adobe adds new AI features to Acrobat and its collaborative Spaces tool, enabling users to generate podcast summaries, create presentations from document content, and edit files using text prompts. This expands its suite of AI capabilities amid growing competition from tools like Canva and NotebookLM.
Googles hybrid AI physics weather model NeuralGCM improves global precipitation forecasts and multi decade simulations by training directly on NASA satellite data It shows significant gains over traditional models in predicting daily cycles averages and extreme rainfall events
Googles hybrid AI physics weather model NeuralGCM improves global precipitation forecasts and multi decade simulations by training directly on NASA satellite data It shows significant gains over traditional models in predicting daily cycles averages and extreme rainfall events
