Hundreds of scientists across more than 30 climate-modelling centres are now running the most ambitious coordinated weather simulation project ever attempted — and the sheer scale of data they're generating is staggering. The seventh phase of the Coupled Model Intercomparison Project, known as CMIP7, will produce more than five million gigabytes of information, so vast that downloading it on a fast internet connection would take two and a half years.
For four decades, CMIP has served as the backbone of global climate science, providing a framework that allows researchers around the world to run the same experiments on their different models and compare results. This standardized approach has been crucial: the projections that emerge from CMIP underpin the landmark assessment reports of the Intergovernmental Panel on Climate Change (IPCC), which in turn guide climate policy decisions affecting billions of people.
Each climate-modelling centre builds its models differently, which means each produces slightly different results — even when fed identical inputs. CMIP solves this puzzle by coordinating common experiments that use the same conditions, letting scientists see where their models agree and where they diverge. The models themselves simulate the physics, chemistry and biology of Earth's atmosphere, land and oceans using supercomputers powerful enough to crunch millions of variables: temperature, rainfall, winds, sea ice extent, ocean currents, and more.
CMIP7 introduces several innovations designed to make this phase more efficient and responsive to urgent climate questions. At its core is the DECK — diagnostic, evaluation and characterisation of klima — a suite of standardized experiments that every modelling centre must run to demonstrate how their model behaves under simple, consistent conditions. These are requested directly by CMIP's scientific governing panel to ensure continuity across decades of research.
But CMIP7 goes further. Working alongside the DECK are 40 different model intercomparison projects (MIPs), each led by specialized research communities investigating particular climate questions at their own pace. Some groups explore what climate might look like under different emission scenarios; others investigate how sea ice has shifted between ice ages. Rather than forcing all research onto a single timeline, these MIPs work with their own schedules, allowing flexibility and depth.
The most significant innovation is the "assessment fast track" — a streamlined subset of CMIP7 experiments identified through community consultation as absolutely critical for science and policy. Data from these fast-track experiments will feed directly into the IPCC's seventh assessment report, scheduled for publication in the coming years. This same data will also support organizations working on regional climate downscaling and modelling everything from ice-sheet changes to climate impacts on specific sectors and regions.
As scientific understanding deepens and computing power improves, each new CMIP phase pushes the boundaries further. CMIP7's models feature higher spatial resolution than ever before, larger ensembles of simulations to better capture uncertainty, more sophisticated representations of key climate processes, and more efficient designs that make computation faster. Across participating countries — from the United States and Europe to China, Japan and beyond — researchers are racing to complete these experiments, knowing their results will shape how the world understands and responds to climate change for the next decade.