OpenAI has released the o3-mini, available today to all users on ChatGPT

Top 3 AI Tokens That May Rise in February 💬

⏺TAO. Over the past month, TAO has fallen by 18% and formed a bottom at $362. Then it began to recover, which highlights the interest in AI tokens. If the upward momentum accelerates, TAO may test the resistance levels at $459 and $495.

⏺GRIFFAIN. The token grew steadily in December, and by January 22, the coin's market cap reached $600 million. But then GRIFFAIN faced a sharp correction. If interest in AI agents returns, the coin may recover above the resistance levels of $0.218 and $0.31.

⏺ARC. The token showed strong growth, reaching a peak capitalization of $622 million. The subsequent correction reduced the total value of all ARC tokens in circulation to $221 million. In case of a reversal, ARC may rise to the resistance levels of $0.279 and $0.348.

OpenAI announced Deep Research, a new ChatGPT model that can conduct autonomous research by searching the web and synthesizing knowledge into a research paper as output, in what is the next step towards AI discovering new knowledge for itself

OpenAI's Deep Research scores a new state-of-the-art score on Humanity's Last Exam, surpassing the accuracy of o3-mini.

Sam Altman says new Deep Research AI agent powered by their o3 model can do "a single-digit percentage of all economically valuable tasks in the world" and this step into synthesizing knowledge will soon be followed by agents that can invent new knowledge.

How will institutions evolve as AI agents enter the workforce? From legacy systems to multi-agent startups, from human quirks to principal-agent efficiency.

ARK Invest has released their highly anticipated "Big Ideas 2025" report, outlining a future where technological convergence drives unprecedented economic growth and innovation.

Here's what you need to know:

The report identifies five key innovation platforms that are evolving and converging:
- AI
- Robotics
- Energy Storage
- Public Blockchains
- Multiomics Sequencing

Economic Impact

ARK projects global GDP growth could reach 7.3% by 2030, significantly higher than the IMF's forecast of 3.1%. Disruptive innovation companies could command more than two-thirds of global equity markets, with market value expanding beyond the current tech giants.

Key Technology Trends

1. AI Agents
- OpenAI projected to surpass $10B in revenue by 2025
- AI-mediated advertising could capture 54% of the $1.1T digital ad market by 2030
- AI agents revolutionizing consumer interaction and business workflows

2. Bitcoin & Digital Assets
- Bitcoin reached new all-time highs in 2024
- Spot Bitcoin ETFs marked the most successful ETF launch in history
- 2030 price targets: $300,000 (bear case) to $1.5M (bull case)
- Stablecoins overtook Mastercard and Visa in transaction value

3. Autonomous Vehicles
- Robotaxis could reduce transport costs to $0.25 per mile
- Market potential of $34T by 2030
- Tesla and Waymo leading commercial deployment

4. Energy & Infrastructure
- AI computing driving unprecedented energy demand
- Nuclear energy gaining renewed attention
- Battery costs continuing to decline
- Renewable energy integration accelerating

5. Robotics & Manufacturing
- Humanoid robots entering commercial use
- $26T+ global revenue opportunity
- 3D printing market growing 40% annually
- Manufacturing becoming increasingly automated

6. Healthcare Innovation
- AI dramatically reducing drug development costs
- DNA sequencing costs falling faster than Moore's Law
- Precision medicine becoming mainstream
- Multi-cancer screening revolutionizing diagnostics

State of AI: China - Key Insights from Q1 2025

While the US maintains an overall lead in the intelligence frontier, China is no longer far behind.

1. Closing the AI Gap

- Chinese AI models have rapidly caught up to US capabilities
- DeepSeek R1 achieved 89 on Intelligence Index, approaching OpenAI's o3 (94)
- Multiple Chinese models now match frontier US model performance

2. Leading Players & Performance
Top Chinese Models:

- DeepSeek R1: 89
- Qwen 2.5 Max (Alibaba): 79
- DeepSeek V3: 80
Market Leaders:
- Big Tech: Alibaba, Baidu, ByteDance, Tencent
- Rising Stars: DeepSeek, MoonShot AI, Zhipu

3. Export Control Impact

- US restrictions on high-end AI chips continue
- NVIDIA H100 (989 TFLOPs) banned for export
- China-approved NVIDIA H20 limited to 148 TFLOPs

4. Investment & Scale
Major Funding:

- MoonShot AI: $1.67B
- Zhipu: $1.12B
- Baichuan: $1.04B

5. Emerging Trends
- Focus on reasoning capabilities
- Open-source model development
- Multiple companies releasing frontier-level models

Google showed a very cool commercial for its Gemini AI chatbot

According to the plot of the video, a man with a Google Pixel 9 in his hands asks Gemini to help him prepare for an interview. Talking about himself, cute family shots from his past appear in the video.

At the end of the video, Gemini says that the hero is ready and the man joins the call with future employers.

By the way, this video will be shown at the Super Bowl, which will be held on the night of February 9th to 10th.

Prime intellect introduced SYNTHETIC-1: Collaboratively generating the largest synthetic dataset of verified reasoning traces for math, coding and science using DeepSeek-R1.

SYNTHETIC-1:

- 1.4 million high-quality tasks & verifiers
- Public synthetic data run - allowing anyone to contribute compute
- GENESYS: open, extendable synthetic data generation framework + call for crowdsourcing tasks & verifiers.

Figure AI to produce over 100,000 humanoid robots by 2029 👋

Figure AI has signed a second contract for the supply of robots of its own production, the client being an unnamed large company from the USA.

Thus, by 2029, Figure AI should produce 100,000 humanoid robots.

The robots are expected to be used to replace physical labor in manufacturing.

We do voiceovers for videos — a model for creating audio from text, Kokoro v1.0, has been released.

Simple but high-quality audio tracks are created. The neural network works locally in the browser, so user data does not leak anywhere.

Test it for free here

A new neural network for image generation has appeared on the network — Loras[.]dev. AI creates anime picchis, logos, icons, sketches, and even tarot cards. It's all free.

HuggingFace released 8GB of high quality reasoning math

They temporarily commandeered the HF cluster to generate 1.2 million reasoning-filled solutions for 500,000 NuminaMath problems using the DeepSeek-R1 model.

This is significant for AI development because:

1. It creates a large dataset of mathematical reasoning examples
2. These solutions can be used to train future AI models
3. It demonstrates the current capabilities of AI in solving complex mathematical problems

Datasets.

OpenAI roadmap update for GPT-4.5 and GPT-5

Sam Altman just dropped major news about the future of their AI development. The company is making a dramatic shift toward unified intelligence with GPT-4.5 and GPT-5.

Key highlights from the announcement:

• GPT-4.5 (codenamed "Orion") will be their final traditional model before a complete system overhaul

• GPT-5 represents a groundbreaking merger of all OpenAI technologies, including o3, marking the end of standalone models

• The most surprising part? Free ChatGPT users will get unlimited access to GPT-5's standard intelligence setting

They're moving toward what they call "magic unified intelligence" - essentially making AI that "just works" without users needing to understand the technical details.

Premium features for Plus and Pro users will include enhanced intelligence levels plus access to voice, canvas, search, and deep research capabilities.

GREEN is a new lightweight neural network designed to how we analyze brain activity

GREEN combines two powerful approaches:

1. Wavelet-based frequency filtering to capture dynamic brain rhythms

2. Riemannian geometry to decode spatial patterns in EEG signals

EEG data holds secrets to brain health, cognitive function, and aging—but traditional methods often miss nuanced signals. GREEN changes the game by:

- Detecting subtle brain activity changes with exceptional sensitivity

- Operating on a low computational budget (ideal for real-world applications!)

- Delivering interpretable outputs to link findings to brain mechanisms.

GitHub.

OpenAI released its Reasoning model best practices

1. Use delimiters for clarity: Use delimiters like markdown, XML tags, and section titles to clearly indicate distinct parts of the input, helping the model interpret different sections appropriately.

2. The guide differentiates between reasoning models (e.g., o1, o3-mini) and GPT models (e.g., GPT-4o).

Reasoning models are built for complex, multi-step tasks—such as planning, detailed document analysis, and visual interpretation—while GPT models focus on speed and cost efficiency for well-defined tasks.

- In practice, reasoning models excel at clarifying ambiguous prompts, extracting key details from extensive unstructured data, and performing multi-step planning or code review.

- They are best used with clear, concise prompts that include explicit constraints and delimiters;

- elaborate chain-of-thought instructions are unnecessary since these models reason internally.

Google DeepMind released a short course on AGI safety

The course offers a concise and accessible introduction to AI alignment problems and our technical & governance approaches, consisting of short recorded talks and exercises (75 minutes total).

DeepSeek introduced CodeI/O, a new method that helps AI learn reasoning patterns hidden in code

Models train to predict inputs and outputs of given code, all while explaining its reasoning with Chain-of-Thought (CoT) in natural language.

CodeI/O improves models' general reasoning skills, such as:

- planning steps logically
- searching for solutions
- breaking problems into smaller parts

DeepSeek gathered over 810,000 Python code files from different sources to cover a big variety of reasoning styles, like puzzles, math problems, etc.

Then they cleaned and structured it into a unified format using DeepSeek-V2.5, ran the code and collected input-output pairs.

CODEI/O++: Improving training with multi-step feedback

It's an improved dataset where the model learns not just from its successes but also from its mistakes. To create it, researchers used a feedback loop:

- If the model gets an answer wrong, it's told why it was wrong and tries again.
- If the function fails to run, researchers include that feedback too.
- The model then revises its response and tries again.

This extra training step makes models even more accurate.

Data and models.