Microsoft released GitHub Spark — a new tool in Copilot that turns your ideas into full-stack apps, entirely in natural language.

MIT introduced MEM1: RL for Memory Consolidation in Long-Horizon Agents.

Long-horizon agents (e.g., deep research, web agents) typically store all observations, actions, and intermediate thoughts in context. However, much of this information is unnecessary for subsequent reasoning, leading to inefficient memory usage and slower inference.

In MEM1, researchers introduced RL approach that trains the agent to maintain a dynamic internal state, which:

1. Consolidates and maintains only relevant information
2. Updates memory while reasoning
3. Discards unneeded history dynamically

A new method achieves:
1. 3.7× lower memory usage & 1.78× faster inference on multi-question HotpotQA
2. 2.5× lower memory usage on WebShop

code and model are fully open-sourced:

Paper.

Chat Annotator— a free chatbot where users can highlight parts of responses, leave a comment, and have the model incorporate that feedback into its next output. Powered by Cohere Command-A.

How do we train LLMs on real-world tasks where it’s hard to define a single verifiable answer?

Scale introduced Rubrics as Rewards (RaR) — a framework for on-policy post-training that uses structured, checklist-style rubrics as interpretable reward signals.

ASI-Arch is the first Artificial Superintelligence for AI Research enabling fully automated neural architecture innovation.

No human-designed search space. No human in the loop.

Key Breakthroughs of ASI-Arch:

- Autonomous code generation & training
- 1,773 experiments conducted (20K+ GPU hours)
- 106 new SOTA linear attention architectures discovered
- Unveiled a scaling law for scientific discovery

Claude Code is getting a brand new feature: custom subagents

Subagents let you create teams of custom agents, each designed to handle specialized tasks.

Examples of subagents we’ve seen be useful are:

1. Software Architect: help design features elegantly and ensure appropriate layers of abstraction.

2. Code reviewer: Review best practices in a codebase, delete old code.

3. QA tester: Run unit tests, lints and writes fixes.

Hunyuan released 3D World Model 1.0

It's the industry's first open-source 3D world generation model, compatible with CG pipelines for full editability & simulation. Set to transform game development, VR, digital content creation and so on.

Try it.

A new world wodel from Meta - DINO-world: a generalist video world model that predicts the future—in latent space.

Trained on uncurated videos with DINOv2, it learns diverse temporal dynamics (driving, indoors, sims), beats prior models on segmentation & depth, and even grasps intuitive physics.

It can be fine-tuned for action-conditioned planning.

Singapore's Sapient Intelligence dropped a Hierarchical Reasoning Model, with a brain-inspired architecture

With training on just 1K examples and 27M params, it handles complex reasoning tasks like extreme Sudoku and maze puzzles

Code.

PayPal to let US. merchants accept payment in more than 100 cryptocurrencies.

The Step-3 tech report is now on arXiv

Step-3 is Large yet Affordable: Model-system Co-design for Cost-effective Decoding.

Chinese lab Z.ai dropped GLM-4.5 and 4.5 Air, 2 open-source agentic models

The 4.5 variant with 355B params tops open models worldwide, and ranks just behind o3 and Grok 4

Also excels at agentic tasks with a 90% success in tool use.

API Pricing (per 1M tokens):
GLM-4.5: $0.6 Input / $2.2 Output
GLM-4.5-Air: $0.2 Input / $1.1 Output

Weights
API
OpenRouter
Develop Tools.
Try them.

OpenAI Introduced study mode in ChatGPT — step by step guidance for students rather than quick answers.

Eigent — the first open source multi-agent workforce on your desktop.

Eigent is a team of AI agents collaborating to complete complex tasks in parallel.

It brings together specialized agents, developer, search, document, multi-modal, each designed to work in parallel and adapt to your needs.

Eigent is built on CamelAI open-source multi-agent infrastructures.

It supports:
- Running parallel tasks
- Custom workers
- Cloud version or "Bring Your Own Key" (BYOK)
- Local model deployment
- Human-in-the-loop feedback
- Model Context Protocol (MCP) tools
- Secure self-hosting
- Enterprise-level security

Eigent supports multiple deployment options:

- Cloud version with instant access and managed infrastructure
- Community edition for local hosting and customization
- Enterprise edition with SLAs, auditability, and scale

Coinbase and JPMorgan have partnered to crypto access for over 80 million Chase customers, introducing three 3 methods:

- converting Chase Ultimate Rewards to USDC,
- funding Coinbase accounts with Chase credit cards,
- direct bank integration.

The integration of Ultimate Rewards to USDC offers a novel entry point, while credit card funding and direct bank links streamline transactions, potentially boosting adoption rates among mainstream users.

BlockDL a free & open-source GUI that lets you visually design Keras neural networks and learn ML.

Langchain introduced Deep Agents

Team created a new Python package which makes it easy to build your own Deep Agents.

The core algorithm for Deep Agents is actually the same - it’s an LLM running in a loop calling tools. The difference is:

1. Planning tool
2. Sub agents
3. File system
4. A detailed system prompt (prompting is not dead!)

Google Introduced AlphaEarth Foundations an AI model that integrates petabytes of satellite data into a single digital representation of Earth.

It'll give scientists a nearly real-time view of the planet to incredible spatial precision, and help with critical issues like food security, deforestation & water resources

Deep cogito released 4 hybrid reasoning models of sizes 70B, 109B MoE, 405B, 671B MoE under open license.

The models are built on Deep cogito’s work on building superintelligence using Iterated Distillation and Amplification (IDA). In particular, team scale the model’s intelligence prior by the model internalizing the reasoning process using iterative policy improvement, rather than simply searching longer at inference time.

This seems to be a novel scaling paradigm where the models develop more “intuition”, and serves as a strong proof of concept for self-improvement. Since the Cogito models develop a better intuition of the trajectory to take while searching at inference time, they have 60% shorter reasoning chains than Deepseek R1.