NVIDIA Cosmos 3: The World's First Fully Open Omnimodel for Physical AI
NVIDIA has announced Cosmos 3 at GTC Taipei COMPUTEX — the world's first fully open omnimodel for Physical AI, combining native vision reasoning, world generation, and action simulation in a single model. Available now in Super (32B) and Nano (8B) variants.
Read more →
MiniMax M3: The First Open-Weight Model with Frontier Coding, 1M Context, and Native Multimodality
MiniMax has released M3, the first open-weight frontier model to combine top-tier coding performance, a 1 million token context window via the new MSA sparse attention architecture, and native multimodality — all in a single model that can also control a desktop computer.
Read more →
Gemini Omni: Google DeepMind's Multimodal Model That Creates Anything From Anything
Google DeepMind has released Gemini Omni, described as their first step toward a model that can create anything from anything — starting with video. It merges Gemini's reasoning intelligence with Google's generative media systems, enabling text, images, audio, and clips to be transformed into video with a conversational interface.
Read more →
mxbai-rerank-v3-listwise: Mixedbread's New Reranker That Goes Beyond Binary Relevance
Mixedbread's mxbai-rerank-v3-listwise reads the whole candidate set at once to resolve conflicts and rank by directives like recency and source priority — delivering +11% NDCG@10 over previous methods.
Read more →
Claude Opus 4.7 Fast Mode Is Now in Research Preview on the API and Claude Code
Anthropic has launched Fast Mode for Claude Opus 4.7 in research preview, available on the API and in Claude Code — bringing significantly lower latency to their most capable model without sacrificing quality.
Read more →
Meta AI Gets Muse Spark Voice Conversations, Live Camera AI, and Real-Time Shopping
Powered by Muse Spark — Meta Superintelligence Labs' first model — Meta AI now supports natural voice conversations with interruptions, live camera understanding, Reels-powered shopping, and real-time image generation.
Read more →
Kimi K2.6: Moonshot AI's Open-Source Model That Swarms Complex Tasks With 1,000 Parallel Agents
Kimi K2.6 is Moonshot AI's open-source 1-trillion-parameter MoE model with long-horizon coding, agent swarm capabilities (up to 1,000 sub-agents), multimodal design, and persistent 24/7 autonomous operation.
Read more →
ModelScope: The Open-Source Model Hub Powering China's AI Ecosystem
ModelScope is Alibaba's open-source model hub and platform hosting thousands of AI models across NLP, vision, audio, and multimodal — serving as the Hugging Face equivalent for China's AI developer community.
Read more →
ChatGPT Launches Personal Finance Feature: Connect Your Bank Accounts and Get AI-Powered Money Insights
OpenAI has rolled out a new personal finance experience inside ChatGPT, letting Pro users in the U.S. securely link their financial accounts and get AI-driven insights grounded in real spending data.
Read more →