Best Science AI Models
5 models tracked · 60 recent news stories
Most-talked-about Science right now
Ranked by mentions across 30+ AI sources in November 2026.
AlphaGenome is Google DeepMind's deep-learning model for predicting how DNA sequences regulate gene expression and how single-letter variants can disrupt that regulation, first released in June 2025. In September 2026 DeepMind published AlphaGenome Atlas, a roughly 1-petabyte precomputed map of the predicted molecular effects of all 9 billion possible single-letter changes in the human genome, with an impact score for every variant.
Carbon-X is Hugging Face's open DNA language model, released on 25 October 2026 under Apache 2.0 — 14.8B parameters built on Qwen3-14B. It is the first model to put DNA and natural language in the same chat: it reads raw nucleotide sequences and answers about them in English, averaging 66.7% across five DNA benchmarks.
Timestral is Mistral AI's time-series forecasting model family, released on 15 October 2026 and spanning 5M to 209M parameters. Mistral reports that the 209M model leads GIFT-Eval, TIME and fev-bench among zero-shot forecasters and beats Google's TimesFM-3 — and, being that small, it runs on a CPU with no GPU needed.
WeatherNext is Google DeepMind's family of machine-learning weather forecasting models, which predict global atmospheric conditions and cyclone tracks up to 15 days ahead. WeatherNext 2 generates each forecast scenario in under a minute on a single TPU — far faster than traditional physics-based numerical weather prediction — and powers weather features across Google Search, Gemini and Pixel.
📰 Latest Science Model News(60 stories)
Google has unveiled AlphaGenome 2 to help identify the causes of diseases
2 sources
Hugging Face launches Carbon-X, an open AI model for natural-language DNA analysis
2 sources
AlphaGenome Atlas Predicts Effects of Billions of DNA Changes
19 sources
Timestral, t0, Tabby, and Morpheus-2.0 Compared on GIFT-Eval
by Sean Moran — All four time-series foundation models scale, patch, and predict quantiles; the real split is how series share information. The post Timestral, t0, Tabby, and Morpheus-2.0 Compared on GIFT-Eval...
Inside Four New Time-Series Foundation Models: How Timestral, t0, Tabby and Morpheus-2.0
by Sean Moran — All four scale a series, cut it into patches, run a transformer and output quantiles. They differ in how series exchange information, how…
2 sources
[Timestral]일정을 전달하면 예측이 바뀔까? 소형 AI를 CPU로 구동해 확인해 보았다
직접 만든 앱의 접속 수를 예측하고 싶다. 하지만 내일은 공지를 올릴 예정이다. 어제까지의 숫자만 AI에게 전달해서 그 변화까지 알 수 있을까?내가 궁금했던 것은 예측 모델의 똑똑함보다, 내가 가지고 있는 '일정'을 어떻게 전달할 것인가였다. 그래서 이번에는 수치의 시계열을 예측하는 Timestral의 소형 버전을 CPU로 구동하여, 일정을 전달하는...
Tigem contributes to Carbon-X, the Open-Source AI Model helping decode the human genome
Timestral: Why Forecasting Tomorrow Is Not the Same as Predicting the Next Token
by Milind Divre — Your chatbot finishes the sentence “Tomorrow’s demand will be…” and confidently produces another sentence. Your operations team wanted a…
With Timestral, Mistral Takes On Austria’s NXAI in Time-Series A.I.
2 sources
Will Artificial Intelligence Make Weather Forecasts More Accurate? — 21st Century Tech Blog
A few months ago, I read that Google had launched an artificial intelligence (AI) weather forecasting app. I thought, good for them… Continue reading on Medium »
