Changelog

Models can now be organized using a nested folder structure, replacing the previous collections system — a more natural fit for how teams build models and manage their work.

What's changed

• Collections have been replaced by folders

• Each model now belongs to a single folder, rather than multiple collections

• Folders can be nested to create a hierarchy that matches how your team works, and how nested models are built

• Existing models have been migrated to the new structure

What's now possible

• Build a folder hierarchy that reflects how your team organizes its work, and how nested models are built

• Navigate and manage models more intuitively

• Lays the groundwork for easier model sharing and collaboration, coming in future releases

Model processes with multiple outputs and define how impacts are distributed across them — enabling more accurate representation of co-products and complex systems.

What’s improved

• Define multiple outputs directly on a process

• Choose which output to calculate against when analyzing results

• Set allocation at the process level using parameters and expressions

• Built-in guardrails ensure allocation sums to 100%

• Upstream allocation handled automatically (no additional setup required)

• Existing single-output models continue to behave as before

What’s now possible

• Model co-products and shared production systems more flexibly

• Explore allocation scenarios by adjusting parameters and expressions

• Analyze results from different output perspectives within the same model

• Better alignment with real-world datasets and more complex LCA workflows

Interacting with your model now happens in side panels instead of blocking popups, so you can keep your model visible while you work.

What’s improved

• Edit parameters alongside your model

• Resize panels to fit your workspace

• Minimize Scenario Analysis and Uncertainty Analysis when you want more modeling space

• Smoother search, sorting, bulk edits, and reordering

• Consistent side panel experience across Parameters, AI Agent, Search, and Analysis

• Clear visual cues when panels are open

• Cleaner layouts and improved responsiveness throughout

What's Now Possible

Less context switching, fewer interruptions, and a faster, more fluid modeling workflow — especially for larger or more complex models.

CarbonGraph now supports full Monte Carlo simulation to quantify and report uncertainty in LCA and EPD results. Read more in the blog.

What's Improved

• Parameter-level probability distributions:

  • Fixed, Normal, Lognormal (μ/σ), Lognormal (Median/GSD), Uniform, Triangular, Beta, Gamma

• Configurable iterations and random seed for reproducible runs

• Default uncertainty handling for unparameterized exchanges

• Full re-evaluation of all impact methods per iteration (TRACI, EF, CML, etc.)

• Results reporting:

  • Percentiles (e.g., P5/P50/P95)

  • Box plots

  • Violin plots

  • Histograms

  • Tables

What's Now Possible

• Reporting statistical ranges alongside impact results

• Sensitivity analysis

• Defensible uncertainty documentation for verification

• More rigorous EPD submissions

• Scientific-grade LCA workflows

Available now in Uncertainty Analysis for any model or flowsheet.

Features

• Added support for exporting model results broken out by EPD category.

Improvements

• Optimized scenario analysis to run faster.

• Made several UI improvements across the platform.

Features

• Added TRACI 2.2 impact method support for ecoinvent and select LCI datasets.

• Added EF 3.1 impact method support for USLCI datasets.