AI2 released OLMo 2, fully open language model to date, including a family of 7B and 13B models trained up to 5T tokens.

OLMo 2 outperforms other fully open models and competes with open-weight models like Llama 3.1 8B.

Model.

Try out an OLMo 2 demo on Playground.

Meta introduced Sparsh is the first general-purpose encoder for vision-based tactile sensing that works across many tactile sensors + many tasks.

The family of models was pre-trained on a large dataset of 460K+ tactile images using SSL.

HuggingFace released SmolVLM, a smol 2B VLM built for on-device inference that outperforms all models at similar GPU RAM usage and tokens throughputs.

SmolVLM can be fine-tuned on a Google collab and be run on a laptop.

Or process millions of documents with a consumer GPU.

Demo.
Model.
Fine-tuning script.

👀Total Stablecoin Value Tops Terra-Era Peak to Near $200 Billion

The total market capitalization for stablecoins has swollen 46% this year to a record of about $190 billion on Sunday, according to DeFiLlama data.

Tether, which issues the world’s largest stablecoin USDT, has seen its token’s circulation soar to nearly $133 billion, accounting for 70% of the total stablecoin market. The company wants to broaden usage of USDT by pushing into new industries including commodities. It recently announced the funding of its first crude oil transaction in the Middle East

Meanwhile, Stripe Inc., the $70 billion fintech giant, in October said it plans to buy stablecoin startup Bridge for $1.1 billion in one of the largest-ever acquisitions of a digital-asset startup. Other traditional finance firms including PayPal Holdings Inc. have also ventured into the sector.

The recent stablecoin surge coincides with a crypto rally spurred by Donald Trump’s victory in the US presidential race. Once an industry skeptic, Trump won over the sector with pledges to put in place more favorable policies.

The overall value of the digital-assets market has risen by nearly a trillion dollars since the election, according to data from CoinGecko.

Alibaba released a preview of QwQ /kwju:/ — an open model designed to advance AI reasoning capabilities.

LLM-brained GUI Agents

Great survey on LLM-brained GUI agents, including techniques and applications.

Mistral announced Startup Program: Mistralship

Become part of a 6-month-cohort of 10 startups to receive:

- 30K credits for La Plateforme
- Dedicated 1:1 support from our Solutions & Science team
- Early access to new models & products

Apply here before January 5th.

First open source reasoning model from Alibaba

Qwen QwQ 32B is in preview, does reasoning like o1 and Deepseek R1 but y'know, on your local machine.

Demo
Model

Scientists at Yale University have developed BrainLM, the first foundation model for analyzing functional MRI brain recordings.

Here's what makes it revolutionary:

- Trained on 6,700 hours of brain activity recordings
- Uses self-supervised masked-prediction training
- Processes data from 77,298 fMRI samples
- Analyzes 424 brain regions simultaneously

Capabilities:

- Accurately predicts clinical variables like age, anxiety, and PTSD
- Forecasts future brain states
- Identifies functional networks without supervision
- Generates interpretable representations of brain activity patterns

What Sets It Apart:

- Generalizes well to new patients and external datasets
- Outperforms baseline models in clinical predictions
- Serves as a powerful "lens" for analyzing massive fMRI repositories
- Creates meaningful insights about brain organization

Potential Applications:
- Non-invasive assessment of cognitive health
- Early detection of psychiatric disorders
- Research tool for understanding brain dynamics
- Biomarker discovery for mental health conditions

Technical Implementation:

- Based on Transformer architecture
- Trained on UK Biobank and Human Connectome Project data
- Uses advanced preprocessing and brain parcellation techniques
- Employs state-of-the-art deep learning methods

PrimeIntellect released INTELLECT-1: open-sourcing the first decentralized trained 10B model

INTELLECT-1 base model & intermediate checkpoints
- Pre-training dataset
- Post-trained instruct models by Arcee ai
- PRIME training framework
- Technical paper with all details.

Decentralized training is one of the most important things to track from an AI policy POV - most efforts to control compute rest on the assumption that strategic AI compute needs to be centralized.

HuggingFace.

Pre-training Dataset.

Post-training datasets by Arcee ai: 1 , 2.

Try out Intellect-1 on.

Tesla unveiled an upgraded hand system for its Optimus humanoid robot, featuring 22 total degrees of freedom

The robot teleoperated by a human, can now catch a ball, with tactile sensing and fine tendon control improvements coming by the end of the year.

BCG has released its insightful AI Maturity Matrix report.

Key findings :

1. The report categorizes economies into three groups: AI Emergents, AI Practitioners & Contenders, and AI Pioneers.

2. AI Emergents are economies in the early stages of AI adoption, requiring strategies to build competitiveness.

3. AI Practitioners & Contenders are economies with gradual or exposed AI adoption, including countries from East Asia, Eastern Europe, and South America.

4. AI Pioneers are the top 5 economies (out of 73) that excel in AI integration, R&D, infrastructure, and talent.

5. The report notes a substantial gap in AI preparedness, with over 70% of nations scoring below the halfway mark in areas like skills, research, and ecosystem development.

6. Global AI expenditure is expected to double, reaching $632 billion by 2028, reflecting technology's central role in future economic strategies.

7. The AI Maturity Index is measured by two indices: AI exposure and AI readiness, evaluated across six dimensions (ASPIRE framework).

8. Economies like Luxembourg and Singapore lead in AI readiness due to their reliance on financial and business services.

9. Developing economies like India can benefit from AI applications in 'agritech' and industrial optimization.

10. The report concludes with recommendations for each economy archetype, urging pioneers to drive global standards, contenders to expand AI applications, and emergents to build foundational infrastructure.

Reverse Thinking Makes LLMs Stronger Reasoners

Shows that training LLMs to learn "reverse thinking" helps to improve performance in commonsense, math, and logical reasoning tasks.

It claims to outperform a standard fine-tuning method trained on 10x more forward reasoning.

World Labs has unveiled an AI system that transforms regular images into fully explorable 3D worlds right in your browser

No complex software needed – just upload an image and step into a three-dimensional universe.

What the system can do:

• Analyzes 2D images and creates realistic 3D geometry
• Fills in unseen parts of the scene
• Lets you freely move around like in a video game
• Supports camera effects (depth of field, dolly zoom)
• Maintains the original artistic style

Perfect for:
• Game designers and developers
• Artists and animators
• Filmmakers
• Architects and designers
• Anyone who wants to bring their ideas to life in 3D

This technology will be available to users soon. Stay tuned – the future of 3D generation is here.

Anthropic announced Fellows Program

This is an exceptional opportunity to join AI safety research, collaborating with leading researchers on one of the world's most pressing problems.

ARIA Research has launched a funding program that aims to revolutionize our understanding of neurological disorders through innovative brain computer interface research.

Key highlights:

The program offers funding up to £500k per proposal, targeting high-potential ideas within the Scalable Neural Interfaces space.

ARIA is seeking groundbreaking approaches that could transform how we interact with and understand the human brain at scale.

What they're looking for:

- Bold, innovative ideas that challenge current assumptions about brain interfaces.

- Research ranging from early-stage exploration to pre-commercial science.

- Projects that could fundamentally change our understanding of what's possible in neurotechnology.

Not included:

- Projects within the Precision Neurotechnologies programme scope
- Conventional ideas that could progress without ARIA support
- Commercial or near-commercial products

ARIA's focus on scalable solutions could lead to breakthrough discoveries in treating neurological disorders.

SpaceX is now likely to be valued about $350B in a new tender offer, $100B MORE than what was discussed last month and would make it the most valuable startup in the world, surpassing Bytedance.

Sakana AI introduced CycleQD: A population-based model merging via Quality Diversity

CycleQD builds on our model merging research, advancing two fronts: evolving a swarm of specialized agents to complement one another, and laying the groundwork for life-long learning by enabling diverse, adaptable skill acquisition at the population-level.

Tsinghua NLP & OpenBMB
introduced the Proactive AI Agent

Even the most advanced AI agents like ChatGPT are still traditional Reactive Agents, requiring explicit user instructions to perform tasks.

But now, new Proactive Agent changes the game. These agents aren’t just instruction followers—they're smart assistants with "insight".

They actively observe, predict human needs, and solve problems before being asked.

Code.

Future House in collaboration with E11 Bio announced a new way to map brain circuits at scale.

The key ingredients are simple (but putting them together has been hard):

1. Electron microscopy, which is standard for connectomics today, is too slow.

2. Standard optical microscopes don't have enough resolution to map brain circuits.

3. Researchers use pan-stains to label membranes and synaptic stains to label synapses.

4. Finally, and crucially: mapping neurons with nanometer resolution across centimeters of axon length, without many any errors, is supremely difficult. Researchers use optical protein barcodes, like brainbow, and multiplexed antibody staining to get the neurons to trace themselves.