AI

if not (WORKING_DIR / “config.json”).exists(): run(qwenpaw_cmd(“init”, “–defaults”), check=False) else: print(“QwenPaw working directory already initialized:”, WORKING_DIR) provider_candidates = [ { “env”: “OPENAI_API_KEY”, “provider_id”: “openai”, “name”: “OpenAI”, “base_url”: ” “model”: os.environ.get(“QWENPAW_MODEL”, “gpt-4o-mini”), “chat_model”: “OpenAIChatModel”, “prefix”: “sk-“, }, { “env”: “OPENROUTER_API_KEY”, “provider_id”: “openrouter”, “name”: “OpenRouter”, “base_url”: ” “model”: os.environ.get(“QWENPAW_MODEL”, “openai/gpt-4o-mini”), “chat_model”: “OpenAIChatModel”, “prefix”: “sk-or-“, }, { “env”: “DASHSCOPE_API_KEY”, … Read more

Anthropic has disabled two of its most capable models for each customer. The shutdown followed a US government order to control exports. The order arrived on June 12, 2026. Named Claude Fable 5 and Claude Mythos 5 directly. Both models were launched just three days earlier, on June 9. The order cited national security authorities, … Read more

This week, Moonshot AI was released Kimi K2.7-Code. It is a coding-oriented, agent-based model. The weight model is submitted to Hugging Face under a modified MIT license. You can also access it through Kimi API and Kimi Code. K2.7-Code targets long-term software engineering, not general discussion. Plans, organizes, implements, and corrects errors in multiple steps. … Read more

The Google Research team announced the launch of the Gemini-SQL2 on X. They described the program as a text-to-SQL breakthrough powered by Gemini 3.1 Pro. Gemini-SQL2 posted an implementation accuracy of 80.04% on the BIRD Text-to-SQL leaderboard (Model One). Google’s chart puts it above its Gemini-SQL, the previous top entry. The metric measures whether the … Read more

Moonshot AI introduced Kimi Work, an AI agent running on your desktop. The Beijing-based AI organization announced it this week along with downloads for macOS and Windows. Kimi Work reads local files, calls your native browser, and performs scheduled tasks. It targets information workers who have problems accessing files and live sessions. Most agent tools … Read more

Zyphra has released Zamba2-VL, a family of open source visual language models. The release includes three sizes: 1.2B, 2.7B, and 7B parameters. Each model is built on a Zamba2 hybrid SSM-Transformer backbone. Visual language models (VLMs) learn images and text together. They answer questions about charts, texts, and pictures. Most open source VLMs use a … Read more

Gemini models have always kept up with the advancements in AI. From text-based chatbots in 2023, Gemini has evolved into a multimodal system capable of understanding and generating text, audio, images… and videos now. AI video production is no longer a standalone tool. With Gemini Omnivideo creation is becoming more common. Gemini Omni this is … Read more

In this tutorial, we build an end-to-end 3D medical image segmentation pipeline using MONAI segmenting the spleen in the Medical Segmentation Decathlon Task09 dataset. We work with volumetric CT scans, apply medical image transformations such as orientation alignment, voxel-spacing normalization, intensity windowing, foreground cropping, and patch-based sampling, and train a 3D NNet model for binary … Read more

Confusion led Deep Research to Computer, its multi-model orchestration program. Improvements improve accuracy, depth of analysis, and quality of citations. Deep Research now breaks down complex questions into smaller tasks and guides them across 20+ parameter models. It returns actionable reports, decks, and dashboards, all within the desktop. Advanced Research in Computing Deep Research is … Read more

Today, xAI launched the Grok Build Plugin Marketplace. It is a built-in catalog of plugins for Grok Build, the company’s code repository agent. The plugin combines capabilities, slash commands, agents, hooks, MCP servers, and LSPs in one package. You browse, install, and update these packages without leaving the terminal. Plugin Marketplace Grok Build is an … Read more