Mentat.fyi
Worth knowing.
A curated map of AI, health, robotics and the megatrends shaping our lives. What's new, what matters, and what might be worth your attention. Explained from the ground up, kept current. Choose your lens.
Overall = breadth + depth + substance + ½·novelty
Quality marks✦Novel▲High-impact⚡Actionable❖Solid✷Hot
Performance of a large language model on the reasoning tasks of a physician
When this study reached peer review, one word vanished from its own title — and the data explains why.
The preprint was titled 'Superhuman performance of a large language model on the reasoning tasks of a physician.' Peer review cut it to 'Performance.'
NEO
China's drug regulator cleared the first brain implant any country has approved for use outside a clinical trial — and it won the race by reading the brain less deeply, not more.
"I couldn't believe I was able to write again" — a patient paralyzed six years, the implant reading intent off the surface of the brain and driving a soft glove over his hand.
The o3 GeoGuessr prompt
The viral prompt that supposedly made OpenAI's o3 superhuman at locating photos had gone the better part of a year without being tested against a plain one. When someone finally ran the control, the one-liner won.
"It'll still be pretty good, except this time it's good because of what you did."
nanoVLM
A model that reads images and answers questions about them, written small enough to read in an afternoon — and then quietly promoted from teaching toy to the bench rig behind a 24-million-sample dataset.
Its makers disown it as a serious model — "not intended to compete" with the best, they wrote, "but rather to demystify" how these systems work. They then ran their real dataset research on it.
Interaction models
Mira Murati's lab put out a model that doesn't wait its turn — it watches you on video, listens, and talks back at the same time, all from one model instead of a stitched-together voice pipeline.
Open systems have done full-duplex audio — interrupting, backchanneling — since Kyutai's Moshi in 2024. The wager that's actually new is the video.
Megatrends
All megatrends →Step into a field for its state of play — what's settled, what's moving, and what's worth watching. Each room is kept live.
Artificial Intelligence
Models, agents, and AI–human collaboration — general-purpose capability scaling into every domain.
Robotics & Physical AI
Humanoids, manipulation, sim-to-real — the hardware catching up to the models.
Longevity & Health
Aging biology, GLP-1, diagnostics, and ML moving into the clinic.
Human Enhancement
BCIs, neural decoding, and cognitive augmentation — closing the loop between mind and machine.
Sub-trends
All →The narrower rooms inside each megatrend — a sharper cut when a field gets too big for one view.
Agents
Models that act, not just answer — tool use and long-horizon reliability.
Frontier & AGI
Scaling laws, timelines, and the takeoff debate.
Open Models
Open weights and local inference — the democratisation track.
Neurotech
BCIs and neural decoding — reading signal straight from the brain.
Discover
Discover all →The players, tools and models that matter — scored on their own merits. Turn the same dial as the feed.
Overall = breadth + depth + substance + ½·novelty
Hugging Face
Open-source platform and model hub for hosting, sharing, and running machine-learning models, datasets, and demos, plus the Transformers library underneath much of the field.
▸ Host and share modelsDeepMind
Google's science + AGI research lab — maker of Genie 3 and a steady stream of frontier work.
OpenAI
AI research and products company behind GPT models, ChatGPT, DALL-E, and Sora; operates as a public benefit corporation.
▸ Large language modelsAnthropic
Frontier AI lab behind Claude — the models and the agent harnesses built on them.
llama.cpp
The low-level inference engine most local-model tools are built on — maximal control, minimal dependencies.
▸ Run a model locallyGemini Omni
Google DeepMind multimodal model that accepts text, images, audio, and video as input and generates short video with synced audio, edited across a multi-turn conversation.
▸ Video generation