In partnership with
Dear Readers,
How close are we to machines that not only calculate, but actually think? In this issue, we take you to the forefront of the AI revolution: Google DeepMind has done it – Gemini Deep Think is the first system to win the gold medal at the International Mathematical Olympiad (IMO). But OpenAI is following suit, proving that the race for general intelligence is finally on. Two strategies, one goal: AI that challenges the human mind.
In addition to this milestone, we take a look at the latest developments in robotics – from Tesla's Optimus to Walker's S2 with self-replacing battery – and at the surprising comeback of Qwen in the model ranking. This issue is full of moments that show how quickly the future becomes the present. Read on and discover how these advances could reshape research, technology, and our everyday lives.
In Today’s Issue:
Two AI giants just achieved what was once thought impossible in mathematics
A new AI inspired by the human brain is acing complex reasoning tests without pretraining
Qwen's new update ditches "hybrid thinking" for a major quality boost
Developers just got a new hackable AI assistant for their terminal
And more AI goodness…
All the best,
Google DeepMind also achieves gold in the IMO!
The Takeaway
👉 Gemini Deep Think solves 5 out of 6 IMO tasks – officially certified at gold level.
👉 OpenAI's model achieves the same score with massive computing power.
👉 Two AI giants are in a neck-and-neck race for generalized reasoning.
👉 Potential: Research, medicine, and technology could benefit greatly from AI-supported thinking.
Google DeepMind has achieved a historic breakthrough with its latest version, Gemini Deep Think: the AI achieved the gold medal standard at the International Mathematical Olympiad (IMO). It solved 5 out of 6 problems in natural language within the regular 4.5-hour time limit – and the result was officially certified by the IMO committee.
This success was made possible by a newly developed parallel reasoning system that pursues multiple hypotheses simultaneously, as well as a special coaching database with IMO solutions. This sets Gemini apart from earlier models such as AlphaGeometry, which still relied on heuristic approaches.
Very excited to share that an advanced version of Gemini Deep Think is the first to have achieved gold-medal level in the International Mathematical Olympiad! 🏆, solving five out of six problems perfectly, as verified by the IMO organizers! It’s been a wild run to lead this
— #Thang Luong (#@lmthang)
4:33 PM • Jul 21, 2025
But OpenAI is not standing still either: with an experimental reasoning model, OpenAI achieved the same score – 5 out of 6 tasks – validated by independent IMO gold medalists. OpenAI relies on massive test time compute and a “reflection loop” to dynamically optimize solution approaches.
Two labs, two strategies – both on track for gold. The question is no longer whether AI will achieve human-level thinking, but when it will revolutionize research and innovation.
Why it matters: Google and OpenAI have achieved IMO gold – a milestone for AI reasoning. Their systems can now solve tasks that were previously reserved for the most brilliant humans.
Source:
Ad
Find out why 1M+ professionals read Superhuman AI daily.
AI won't take over the world. People who know how to use AI will.
Here's how to stay ahead with AI:
Sign up for Superhuman AI. The AI newsletter read by 1M+ pros.
Master AI tools, tutorials, and news in just 3 minutes a day.
Become 10X more productive using AI.
In The News
🚀Introducing Hierarchical Reasoning Model🧠🤖
Inspired by brain's hierarchical processing, HRM delivers unprecedented reasoning power on complex tasks like ARC-AGI and expert-level Sudoku using just 1k examples, no pretraining or CoT!
Unlock next AI breakthrough with
— #Guan Wang (#@makingAGI)
1:23 PM • Jul 21, 2025
Qwen Releases New Instruct Model
Qwen has released an updated version of its powerful 235B model, which abandons the previous "hybrid thinking" approach in favor of training separate "Instruct" and "Thinking" models for maximum quality.
Grok for Your Terminal
A new, hackable, open-source AI agent called Grok CLI has been released, bringing the power of Grok directly into the command line without relying on complex LLM frameworks.
Graph of the Day
A new iteration of Qwen achieves top marks and even beats the recently released Kimi-k2, which caused quite a stir.
The development of robotics continues to advance inexorably. Figure.02 has now developed a new type of battery that performs significantly better.
Today we're introducing our next generation humanoid battery for F.03
Like our actuators, vertically integrating our battery system is critical to Figure’s success
Engineered in-house and manufactured at BotQ
— #Figure (#@Figure_robot)
3:41 PM • Jul 17, 2025
The Chinese robot Walker S2 has now achieved something truly remarkable: it can replace its own battery. This enables it to work for 24 hours without needing to be recharged.
Walker S2 replacing it's own battery
24/7 blue-collar-worker within reach
— #Chubby♨️ (#@kimmonismus)
8:56 PM • Jul 18, 2025
Tesla Optimus is being used more and more widely. Presumably still teleoperated, but it makes a very big impression.
Humanoids make so much sense. No need to redesign the world or make purpose-built tools.
When not serving popcorn, it can potentially operate and maintain the popcorn machine or even clean up the floor.
Probably teleop, but a great use case to target.
— #The Humanoid Hub (#@TheHumanoidHub)
11:08 PM • Jul 21, 2025
Share Your AI & Robotics Innovation with 200,000+ Readers
Are you building the future of robotics powered by AI? We’re featuring projects at the intersection of artificial intelligence and robotics in Superintelligence, the leading AI newsletter with 200k+ readers. If you have exciting product presentations about your robotics products or significant breakthroughs, please show them to us.
Submit your research or a summary to [email protected] with the subject line “Robotics Submission”. If selected, we’ll contact you about a possible feature.
Question of the Day
Quote of the Day
Sponsored By Vireel.com
Vireel is the easiest way to get thousands or even millions of eyeballs on your product. Generate 100's of ads from proven formulas in minutes. It’s like having an army of influencers in your pocket, starting at just $3 per viral video.
