Editor’s Pick
You should not miss this one
[Open Source Thinking Model] Moonshot AI Releases Kimi K2 Thinking: An Impressive Thinking Model that can Execute up to 200–300 Sequential Tool Calls without Human Interference. Kimi K2 Thinking is an open weights thinking agent built on a 1T parameter Mixture of Experts with 32B active parameters, 256K context window, and Multi head Latent Attention, optimized for test time scaling with hundreds of tool calls per task and large thinking token budgets on benchmarks such as Humanity’s Last Exam, BrowseComp, and SWE bench with tools. It runs with native INT4 quantization and Quantization Aware Training, making long horizon reasoning and tool use practical for real systems.
AI Dev and Latest Releases
[Agentic AI] CMU Researchers Introduce PPP and UserVille To Train Proactive And Personalized LLM Agents. The research team introduces UserVille, an interaction centric environment with LLM based user simulators, and PPP, a multi objective RL framework that optimizes Productivity, Proactivity, and Personalization together. Using Seed OSS 36B Instruct with GRPO and DAPO style token level updates, PPP trains agents to ask low effort clarifying questions and follow user preferences. On SWE Bench Verified function localization and BrowseComp Plus with vague prompts, PPP improves the base model by an average of 16.72 points and outperforms GPT 5 by 21.6 on average.
[Agent for Data Science] Google AI Introduces DS STAR: A Multi Agent Data Science System That Plans, Codes And Verifies End To End Analytics. DS STAR shows that practical data science automation needs explicit structure around large language models, not only better prompts. The combination of Aanalyzer, Averifier, Arouter and Adebugger turns free form data lakes into a controlled Text to Python loop that is measurable on DABStep, KramaBench and DA Code, and portable across Gemini 2.5 Pro and GPT 5. This work moves data agents from toy table demos toward benchmarked, end to end analytics systems.
[AI Agent with GO] Google AI Releases ADK Go: A New Open-Source Toolkit Designed to Empower Go Developers to Build Powerful AI Agents. Google’s Agent Development Kit now has an idiomatic Go SDK, so you can build AI agents directly inside existing Go services using the same framework that powers Python and Java agents. The tutorial walks through creating a small “time in city” agent that calls Gemini with gemini-2.5-flash, wires in the Google Search tool, and runs both in a terminal and the browser based dev UI, giving early level AI engineers a reproducible pattern for production style agent development in Go.
[Open Source Foundation Model] Prior Labs Releases TabPFN-2.5: The Latest Version of TabPFN that Unlocks Scale and Speed for Tabular Foundation Models. TabPFN 2.5 is Prior Labs’ new tabular foundation model that scales in context learning to datasets with up to 50,000 samples and 2,000 features, delivers state of the art accuracy on TabArena and internal industry benchmarks, and matches AutoGluon 1.4 without any per dataset tuning, by using a pretrained transformer trained on synthetic tabular tasks, a Real TabPFN 2.5 variant fine tuned on real data, a distillation engine for low latency MLP or TreeEns students, and a non commercial tabpfn 2.5 license for research and internal evaluation.
Project Notebooks/Tutorials
▶ Build a Multi-Agent System for Integrated Transcriptomic, Proteomic, and Metabolomic Data Interpretation with Pathway Reasoning Codes Tutorial
▶ How to Build a Model-Native Agent That Learns Internal Planning, Memory, and Multi-Tool Reasoning Through End-to-End Reinforcement Learning Codes Tutorial
▶ Build an Autonomous Wet-Lab Protocol Planner and Validator Using Salesforce CodeGen for Agentic Experiment Design and Safety Optimization Codes Tutorial
▶ How to Design a Persistent Memory and Personalized Agentic AI System with Decay and Self-Evaluation? Codes Tutorial
▶ How Can We Build Scalable and Reproducible Machine Learning Experiment Pipelines Using Meta Research Hydra? Codes Tutorial
▶ How to Build an Advanced Multi-Page Reflex Web Application with Real-Time Database, Dynamic State Management, and Reactive UI Codes Tutorial
▶ How to Build an Agentic Decision-Tree RAG System with Intelligent Query Routing, Self-Checking, and Iterative Refinement? Codes Tutorial