Google will release its first fully self-developed smartphone chip, Tensor G5, in the upcoming Pixel 10 smartphone to be unveiled 8/20, media report, a break from the past when Google worked with Samsung on the chip.

Google also switched manufacturers, tapping TSMC’s 3nm process for the Tensor G5.

Apple introduced Foundation Models and a new Foundation Models framework, which gives app developers direct access to the on-device AFM model.

AWS launched Kiro, a new agentic IDE to take on Cursor

It combines agentic coding with spec-driven development to bridge the gap between AI prototypes and complex production-ready apps.

Free preview now available.

Ai2 introduced AutoDS—an AI that doesn’t just hunt for answers, it decides which questions are worth asking.

AutoDS spins up its own hypotheses, runs the stats, learns from the outcomes, and then repeats. The system can use the results of statistical experiments it generates and conducts to propose new hypotheses.

Paper.

Olympiad math + AI: Google ran Gemini 2.5 Pro on the fresh IMO 2025 problems.

With careful prompting and pipeline design, it solved 5 out of 6 — remarkable for tasks demanding deep insight and creativity.

The model could win gold.

Kimi K2 paper dropped. Some notes:

1. MuonClip optimizer.

2. large-scale agentic data synthesis pipeline that systematically generates tool-use demonstrations via simulated and real-world environments.

3. an RL framework that combines RLVR with a self-
critique rubric reward mechanism that allows model to evaluate its own outputs.

IBM introduced a Framework for Quantum Advantage

Researchers defined quantum advantage as a task where results are verifiable and quantum outperforms classical in cost, efficiency, or accuracy.

The pathways forward are methods that are either:
- Provably bounded
- Variational
- Classically verifiable

Paper.

Google DeepMind shared pre-print AMIE research diagnostic dialogue AI.

Researchers introduced a new asynchronous oversight paradigm, decoupling history-taking by AMIE from sharing a human-approved diagnosis.

AMIE can perform consultations with patients to gather information within guardrails (g-AMIE), abstaining from individualized medical advice. A diagnosis and treatment plan is proposed, which licensed physicians authorize through our interface, the clinician cockpit.

Guardrailed-AMIE multi-agent system consists of a multi-phase dialogue agent, a guardrail agent and a SOAP note generation agent based on Gemini 2.0 Flash.

Researchers evaluate workflow in a virtual Objective Structured Clinical Examination (OSCE) study with oversight, contextualizing g-AMIE’s performance with control groups consisting of primary care physicians (PCPs) and nurse practitioners (NPs)/physician assistants/associates (PAs).

g-AMIE and the control groups (g-PCP and g-NP/PA) all operate under the same guardrails of not providing individualized medical advice during consultations and draft SOAP notes for handoff.

This work has various limitations and nuances, including the difficulty of classifying individualized medical advice, the AI-focused workflow which was unfamiliar to both control groups, high mental load required for oversight and the simulated nature of our OSCE study.

Because of this, results need to be interpreted with care and cannot be used to draw conclusions about the relative performance of our PCP, NP and PA control groups.

Alibaba released Qwen3-Coder

This 480B-parameter Mixture-of-Experts model (35B active) natively supports 256K context and scales to 1M context with extrapolation.

It achieves top-tier performance across multiple agentic coding benchmarks among open models, including SWE-bench-Verified.

Alongside the model, also open-sourcing a command-line tool for agentic coding: Qwen Code. Forked from Gemini Code, it includes custom prompts and function call protocols to fully unlock Qwen3-Coder’s capabilities.

Kaggle launched an LLM eval product

How can we teach embodied agents to think before they act?

ThinkAct — a hierarchical Reasoning VLA framework with an MLLM for complex, slow reasoning and an action expert for fast, grounded execution.

Paper.

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.