Meta introduced Code World Model (CWM), a 32B-parameter research model designed to explore how world models can transform code generation and reasoning about code.

Researchers trained the whole stack. 𝗣𝗿𝗲𝘁𝗿𝗮𝗶𝗻. 𝗦𝗙𝗧. 𝗥𝗟. 𝗢𝗽𝗲𝗻 𝘄𝗲𝗶𝗴𝗵𝘁𝘀. 𝗢𝗽𝗲𝗻 𝗺𝗲𝘁𝗵𝗼𝗱𝘀. 𝗢𝗽𝗲𝗻 𝘀𝗰𝗶𝗲𝗻𝗰𝗲.

From tokens to traces. From guesses to grounded.

Code World Model: producing code by imagining the effect of executing instructions and planning instructions that produce the desired effect.

Open weights.

GitHub.

Sakana AI introduced ShinkaEvolve: An open-source framework that evolves programs for scientific discovery with unprecedented sample-efficiency.

Like AlphaEvolve and its variants, framework leverages LLMs to find SOTA solutions to complex problems, but using orders of magnitude fewer resources.

On the classic circle packing optimization problem, ShinkaEvolve discovered a new SOTA using only 150 samples. This is a big leap in efficiency compared to previous methods that required thousands of evaluations.

Researchers applied ShinkaEvolve to a diverse set of hard problems with real-world applications:

1. AIME Math Reasoning: It evolved sophisticated agentic scaffolds that significantly outperform strong baselines, discovering an entire Pareto frontier of solutions trading performance for efficiency.

2. Competitive Programming: On ALE-Bench (a benchmark for NP-Hard optimization problems), ShinkaEvolve took the best existing agent's solutions and improved them, turning a 5th place solution on one task into a 2nd place leaderboard rank in a competitive programming competition.

3. LLM Training: researchers even turned ShinkaEvolve inward to improve LLMs themselves. It tackled the open challenge of designing load balancing losses for Mixture-of-Experts models. It discovered a novel loss function that leads to better expert specialization and consistently improves model performance and perplexity.

ShinkaEvolve achieves its remarkable sample-efficiency through three key innovations that work together:

1) an adaptive parent sampling strategy to balance exploration and exploitation,

2) novelty-based rejection filtering to avoid redundant work,

3) a bandit-based LLM ensemble that dynamically picks the best model for the job.

Paper.
GitHub.

Citi group just dropped new report: Stablecoins 2030 Web3 to Wall Street

Key Takeaways

1. In Citi’s April 2025 Citi GPS: Digital Dollars, we argued 2025 would be a ChatGPT moment for the institutional adoption of blockchain.
The past six months have confirmed this, with digitally native companies leading the “real world” charge.

2. Reflecting rapid YTD growth and new project announcements, we revise our 2030 base case estimate for stablecoin issuance to $1.9 trillion (previously $1.6 trillion) and bull case to $4.0 trillion (previously $3.7 trillion).

3. At 50x velocity (see page 24), similar to fiat payment velocity over time, stablecoins could support nearly $100 trillion in transaction activity by 2030 (base case).The same velocity for our bull case (market size $4.0 trillion) would imply $200 trillion.

4. Citi sees an ecosystem where stablecoins, tokenized deposits, deposit tokens, and CBDCs can all flourish and co-exist.Different forms of money will find different product market fit with usage shaped by trust, interoperability, and regulatory clarity.

5. Bank tokens (tokenized deposits, deposit tokens, and similar), offering the trust, familiarity, and regulatory safeguards of bank money, are preferred by many corporates.2030 bank token transaction volumes could exceed stablecoins.

6. Large corporate treasuries are interested in programmability, enabling real-time settlement/reconciliation, and compliance embedded at the point of transaction, with fewer friction points.These can be offered by bank tokens and stablecoins.

7. On-chain money volumes are likely to remain heavily USD denominated and a source of incremental new demand for U.S. treasuries.
However, Hong Kong, UAE, and other innovative hubs are also hot spots of activity – this is not just about USD.

8. While citi’s annual turnover forecasts for stablecoins (base case: $100 trillion) and bank tokens (above $100 trillion) may appear large to the layperson, this is still small relative to money flows: leading banks move $5–10 trillion per day today.

Key Figures:

- $1.9 trillion is the bank’s revised base case estimate for 2030 stablecoin issuance.

- 0.5 Stablecoin institutional adoption, on a scale of zero to ten.

- $100–140 trillion Estimated transaction volume using bank tokens (tokenized deposits, deposit tokens, hybrids) by 2030.

Arc institute & Stanford introduced Germinal, a generative AI system for antibody design. Fully open source. Good Q&A with the authors below.

Germinal produces functional nanobodies in just dozens of tests, making custom antibody design more accessible than ever before.

Germinal works by integrating AlphaFold-Multimer, which predicts protein–protein structures, with IgLM, an antibody-specific language model.

The result is computational designs that behave like true antibodies.

Apple is working on protein folding and introduced SimpleFold: Folding Proteins is Simpler than You Think

SimpleFold is the first flow-matching based protein folding model that solely uses general purpose transformer blocks.

Protein folding models typically employ computationally expensive modules involving triangular updates, explicit pair representations or multiple training objectives curated for this specific domain.

Instead, SimpleFold employs standard transformer blocks with adaptive layers and is trained via a generative flow-matching objective with an additional structural term.

GitHub

Google released SOTA Gemini Robotics 1.5 model, so it can understand & reason about the physical world.

Gemini Robotics 1.5 is a levelled up agentic system that can reason better, plan ahead, use digital tools such as Google Search, interact with humans and much more.

New capabilities include:

-Powerful spatial reasoning
-Advanced agentic behaviors
-Flexible thinking budget
-Improved safety filters
-And more

Docs.

Google's Startup Technical Guide to AI Agents

A practical framework for founders and executives:

- Architectures for multi-agent systems beyond single-model chatbots
- Integration into product development, operations, and go-to-market functions
- Case studies demonstrating efficiency gains and faster iteration cycles

It's a great read.

Cloudflare introduced NET Dollar, a new $ backed stablecoin that will enable instant, secure transactions for the agentic web.

Cloudflare hosts 20% of the world’s online traffic.

We are said to be entering an age where ai becomes ui and most websites will be crawling with agentic flows and execution.

And in that world, cloudflare having their own stablecoin be the one that agents on those sites use via x402 is very promising for them as a revenue line (see how much money tether makes as a proxy).

Cloudflare also cofounded the x402 foundation alongside coinbase, x402 is an agentic payments protocol.

#DeepSeek introduced DeepSeek-V3.2-Exp — latest experimental model

Built on V3.1-Terminus, it debuts DeepSeek Sparse Attention(DSA) for faster, more efficient training & inference on long context.
Now live on App, Web, and API.
API prices cut by 50%+

DSA achieves fine-grained sparse attention with minimal impact on output quality — boosting long-context performance & reducing compute cost.

Benchmarks show V3.2-Exp performs on par with V3.1-Terminus.

DeepSeek API prices drop 50%+, effective immediately.

For comparison testing, V3.1-Terminus remains available via a temporary API until Oct 15th, 2025, 15:59 (UTC Time).

Key GPU kernels in TileLang & CUDA (use TileLang for rapid research prototyping).

Hf.

Microsoft is launching the next progression of vibe coding this morning with what they are calling 'Vibe Working' in Copilot. It includes:
- Agent Mode in Excel
- Agent Mode in Word 
- Office Agent in chat (this one is powered by Claude)

OpenAI introduced Instant Checkout in ChatGPT with Etsy and Shopify, and open-sourcing the Agentic Commerce Protocol that powers it, built with Stripe, so more merchants and developers can integrate agentic checkout.

Instant Checkout is now rolling out to US ChatGPT Pro, Plus and Free logged-in users buying from US Etsy sellers, with over 1 million Shopify merchants coming soon.

Merchants interested in joining can learn more and apply here.

Google presented ReasoningBank: memory for self-evolving LLM agents

• Distills strategies from both successes & failures
• Enables agents to learn, reuse, and improve over time
• Outperforms prior memory methods on web & SWE tasks (+34.2% eff., –16% steps)

Stripe released the Agentic Commerce Protocol, codeveloped by Stripe and OpenAI.

Also Stripe launched an API for agentic payments, called Shared Payment Tokens.

OpenAI launched a new app called Sora. This is a combination of a new model called Sora 2, and a new product that makes it easy to create, share, and view videos.

Sora 2 can do things that are exceptionally difficult for prior video generation models.

It’s more physically accurate and realistic than prior systems and a big leap forward in controllability. And it also comes with synchronized audio.

There are two ways to access & use Sora 2:

1. The Sora App

The Sora iOS app is available to download now but access is invite-only.

You can sign up in-app for a push notification when access opens for your account.

2. Once you have access to the Sora app, you’ll also be able to access Sora 2 through sora.com.

Starting the initial rollout in the U.S. and Canada today with the intent to expand to additional countries.

Android users will be able to access Sora 2 via sora.com once you have an invite code from someone who already has access.

also plan to release Sora 2 in the API.

And Sora 1 Turbo will remain available, and everything you’ve created will continue to live in your sora.com library.

New from Anthropic: context engineering for AI agents

Anthropic recently published a technical overview of context engineering - managing what information gets fed to language models during execution. This shifts focus from pure prompt design to thinking holistically about the entire information state available to an agent.

The core problem
Language models have finite attention budgets. As you add more tokens to the context window, retrieval and reasoning performance gradually degrades. This happens because transformers create n² token relationships - as context grows, the model's capacity to maintain these relationships gets stretched thin.

Context is a limited resource with diminishing returns.

Key principles

System prompts: Clear and specific, but not so prescriptive they hardcode brittle logic. Find the right level of abstraction between vague guidance and micromanagement.
Tools: Self-contained with minimal overlap. If you can't definitively say which tool applies in a situation, the agent won't do better.
Examples: Curate a small set of diverse examples rather than exhaustively listing edge cases. Most token-efficient way to communicate expected behavior.
General rule: Find the minimal set of high-signal tokens that maximize likelihood of your desired outcome.
Just-in-time retrieval
Instead of pre-loading all potentially relevant data, modern agents maintain lightweight references (file paths, queries, URLs) and dynamically load information at runtime using tools.
This mirrors human cognition - we create indexing systems and retrieve on demand rather than memorizing everything. The tradeoff is speed versus context efficiency. Many effective agents use hybrid approaches.
Long-horizon techniques
When tasks exceed the context window:
Compaction: Summarize conversation history and start fresh. The challenge is choosing what to keep versus discard.
Structured note-taking: Agent maintains persistent notes outside the context window, retrieving them as needed. Works like keeping a TODO list.
Sub-agent architectures: Specialized agents handle focused tasks in clean context windows, returning condensed summaries to a coordinating agent.
Choice depends on task characteristics. Compaction maintains conversational flow. Note-taking suits iterative development. Multi-agent works for complex research requiring parallel exploration.

Epoch AI introduced new AI Companies Data Hub

Researchers collected key data on frontier AI companies, including revenue run rates, funding, staff, usage rates, and compute spend.

Revenue:

The combined revenue rates of OpenAI and Anthropic have grown around 10x since early 2024.

OpenAI’s annualized revenue reached $13 billion in August 2025, up from $5B at the start of the year.

Anthropic’s revenue has exploded this year, from $1B to $5B by July.

Funding:

OpenAI, Anthropic, and xAI have attracted massive investor interest.

OpenAI's last raised at a value of $300 billion, with a $500B valuation under discussion.

And collectively, the frontier AI companies in our data have raised ~$100B in equity and debt funding.

Usage:

The user bases for leading chat applications like ChatGPT and Gemini have continued to grow rapidly.

ChatGPT alone surpassed 700 million weekly active users by August 2025, processing over 3 billion daily messages.

Staff:

OpenAI and Anthropic have both expanded from small startups to thousands of full-time staff, though they are still well behind Google’s flagship AI effort, Google DeepMind.

Compute spend:

Compute for research, training, and inference is expensive: OpenAI’s cloud compute bill for 2025 will exceed $15 billion!

The extensive large-scale data center buildouts underway suggest this rapid growth could continue in the coming years.

E11Bio announced PRISM, a new, scalable technology for mapping brain circuits

PRISM uses molecular ID codes and AI to help neurons trace themselves.

Researchers discovered a new cell barcoding approach exceeding comparable methods by more than 750x.

This is the heart of PRISM. Researchers integrated this capability with microscopy and AI image analysis to automatically trace neurons at high resolution and annotate them with molecular features.

This is a key advance towards economically viable brain mapping - 95% of costs stem from neuron tracing. It is also an important step towards democratizing neuron tracing for everyday neuroscience.

Solving these problems is critical for curing brain disorders, building safer and human-like AI, and even simulating brain function.

In first pilot study, researchers acquired a unique dataset in mouse hippocampus. Barcodes improved the accuracy of tracing genetically labelled neurons by 8x – with a clear path to 100x or more.

They also permit tracing across spatial gaps – essential for mitigating tissue section loss in whole-brain scaling.

Addgene constructs.
Volara.
Open data.

Ex-OpenAI team - Thinking machines introduced Tinker: a flexible API for fine-tuning language models.

Write training loops in Python on your laptop; will run them on distributed GPUs.

Private beta starts today.

Microsoft introduced Agent Framework

You can build, orchestrate, and scale multi-agent systems in Azure AI Foundry using this framework.

Meta Superintelligence labs introduced MENLO: From Preferences to Proficiency

Team introduced a framework + dataset for evaluating and modeling native-like LLM response quality across 47 languages, inspired by audience design principles.

Data.