Skip to main content
​
Faster data uploads, MSMR extrapolation, and performance improvements

​
Faster data uploads

Data uploads now use signed URLs instead of JSON request bodies, eliminating timeouts when uploading large datasets. This is especially noticeable for files that previously failed to upload due to size limits.Find out more →

​
OCP data interpolation with MSMR extrapolation

OCP fitting now blends measured data with MSMR physics-based fits, providing smooth interpolation within the data range and physically-grounded extrapolation outside it.

​
Performance improvements

Pipeline loading is significantly faster thanks to reduced unnecessary fetching and elimination of duplicate element requests. Database queries have also been optimized with new indexes and a more efficient authentication check.
Improvements
  • Default model constants are now available for parameterized models.
  • API responses include a correlation ID for easier troubleshooting.
  • Route guards added for projects and cyclers to prevent unauthorized access.
Fixes
  • Fixed infinite recursion when switching organizations.
  • Fixed variable names in the optimization creation form.
Improvements
  • Composite MSMR support for multi-particle electrode models.
  • Improved round-trip serialization support for optimizers.
Fixes
  • Pchip interpolation errors are now wrapped as NumericalErrors to prevent unexpected pipeline failures.
Improvements
  • Additional solver options can now be passed through as keyword arguments for finer control over solver behavior.
Fixes
  • Fixed solver warnings during protocol simulation.
​
Cell metadata, gzip compression, and improved AI chat

​
Cell metadata support

Cells now support rich metadata with a structured spec format, letting you attach and query detailed cell specifications alongside measurement data.Find out more →

​
Gzip compression for faster uploads

JSON file uploads to storage now use gzip compression, significantly reducing upload times for large parameter files and simulation results.

​
Enhanced AI project chat

The AI chat assistant has been updated with improved simulation awareness, providing more relevant answers to questions about your project data and pipeline results.
Improvements
  • Automatic JSON schema generation for frontend form validation.
  • Removed old parameterization UI in favor of the new pipeline-based workflow.
Fixes
  • Fixed missing .gz extension for child job parameter and metadata files.
  • Fixed pipeline list page sending unnecessary API requests.
  • Improved stale job and pipeline detection reliability.
Improvements
  • Metric weight renamed to action weight for clearer optimization configuration.
  • Added variable support to the design optimization schema.
Fixes
  • Fixed current termination sign handling for charge and discharge steps.
​
AI project assistant, simplified experiment configuration, and initial SOC from UCP

​
AI assistant for projects

Studio now includes an AI-powered chat assistant within each project. Ask questions about your project data, get guidance on setting up simulations and optimizations, and explore results — all from a conversational interface.

​
Simplified experiment and model configuration

Experiment parameters and model configuration have been streamlined with a simpler API. Additional simulation variables are now accessible through the API, making it easier to inspect and customize simulation outputs.
Improvements
  • DataFit job timeout extended to 24 hours.
Fixes
  • Fixed parameterized model variable name lookup for list-type domains.
Improvements
  • Design optimization now reads initial SOC directly from UCP files.
  • Metric values are saved before action execution for better progress tracking.
Improvements
  • Added missing operators to step termination conditions for broader protocol support.
  • Updated compatibility with latest Studio API changes.
Fixes
  • Fixed NumPy compatibility issue.
​
Live progress updates, modernized PSO optimizer, and lumped models

​
Live progress updates during pipeline runs

Pipeline runs now stream live progress updates to the UI, so you can monitor optimization and data fitting progress in real time. Metadata updates are deferred during execution to keep things fast, and pipeline job timeout has been extended to 24 hours.

​
Modernized PSO optimizer

The Particle Swarm Optimization implementation has been modernized with configurable boundary handling options, inertia weighting, constriction coefficients, and velocity clamping — giving you finer control over optimizer behavior.

​
Lumped models and voltage component plots

Studio now supports lumped battery models alongside the existing physics-based models, and voltage component plots are available for visualizing individual voltage contributions during simulation.
Improvements
  • New API endpoints for retrieving optimization variable names.
  • Design optimization schema integrated with the data fitting framework.
  • Validation results moved to metadata storage for improved reliability.
Fixes
  • Fixed experiment protocol format string that could crash the app.
  • Fixed charge optimization regression with new pipeline; prevented accidental baseline data overwrite.
  • Fixed memory issue in pipeline updates.
Improvements
  • Added protocol YAML string support for experiments in design optimization.
  • Added regularizer parsing for optimization constraints.
  • Added visual validation reports for inspecting fit quality.
  • Improved interpolant slope formulation for smoother parameter interpolation.
  • Reinstated reformulate Xj option for electrode models.
  • Updated LGM50 demo notebooks.
Fixes
  • Fixed circular reference memory bugs during multistart optimization.
Improvements
  • Support for custom models with Voltage and Power operating modes, plus improved EIS simulation.
  • Added a converter for translating protocols into PyBaMM experiments.
  • Improved input reference handling when converting protocols to PyBaMM experiments.
​
Redesigned design optimization API and voltage components plot

​
Redesigned design optimization API

Design optimization has been restructured with a cleaner API for configuring models, parameters, experiments, and objectives. A new strict schema enforces valid configurations and enables automatic frontend form generation, making it easier to set up and validate optimization runs.Find out more →

​
Voltage components plot

A new voltage components plot lets you visualize the individual contributions to cell voltage — such as open-circuit voltage, ohmic losses, and reaction overpotentials — for deeper insight into cell behavior during simulations.
Fixes
  • Fixed cells from other projects appearing in cell selection lists.
  • Empty simulation data now correctly shows a failed status instead of a blank result.
Fixes
  • Fixed reaction overpotential (eta_r) calculation bug.
​
Data management for cells, silicon design optimization, and UCP solution save/load

​
Comprehensive data management

Studio now provides a full data management experience for cell specifications, cell instances, and measurements. You can define cell specs, track individual cell instances, and associate measurements — all from a redesigned UI with streamlined navigation.Find out more →

​
Silicon design optimization

The pipeline now supports silicon-based electrode design optimization, including cell mass calculations. This enables end-to-end design studies for cells with silicon anodes.

​
UCP solution save and load

Protocol simulation results can now be saved to disk and reloaded later with lazy variable evaluation. This makes it practical to work with large simulation outputs without keeping everything in memory.Find out more →
Fixes
  • Fixed a bug where some calculation results were not correctly parsed back into the data.
  • Improved error handling for MSMR fits so numerical issues surface clearly instead of failing silently.
  • Cleaned up solver output for a clearer log during fits.
​
EIS simulations, child job status, and saved validation plot config

​
EIS simulations

EIS (Electrochemical Impedance Spectroscopy) simulations now work out of the box — default double-layer capacity parameters are automatically included so you no longer need to manually specify them.Find out more →

​
Child job status visibility

Pipeline and optimization pages now display the status of individual child jobs, making it easy to monitor progress and identify which steps have completed or failed.

​
Saved validation plot configuration

Validation plot settings are now persisted, so you no longer need to reconfigure plots each time you revisit a validation result. This also avoids unnecessary recomputation.
Fixes
  • Fixed authentication configuration and error handling; improved metric display.
  • Fixed datafit parameter bounds format (now uses [lower, upper]).
  • Fixed model selection in optimization, charge time display, and cycler protocol metrics.
Improvements
  • OCP data loading now gracefully ignores steps instead of erroring.
  • Added support for converting fitted costs back into their configuration.
  • Added current to the lumped SPMR model.
Fixes
  • Fixed a bug with custom parameters in pipeline evaluation.
Fixes
  • Fixed step label casing to match standard naming conventions.
​
Models management, data fit performance, and parallel DifferentialEvolution

​
Models management

Studio now has dedicated pages for managing models — browse a list of available models, view model details, and create new models. Parameterized model details are also viewable from the model pages.Find out more →

​
Data fit performance improvements

The data fit UI is now significantly more responsive. React memoization and element collapsing prevent unnecessary re-renders, and payload sizes have been reduced for batch job updates.
Fixes
  • Fixed batch payload format for nested dictionaries.
  • Fixed pipeline data fit element fetching.
Improvements
  • Penalties and constraints refactored to use PyBaMM expressions for greater flexibility.
  • SciPy DifferentialEvolution optimizer now supports parallelization.
Fixes
  • Fixed MSMR validation plot range clipping and a type-conversion bug with large integers.
  • Fixed bugs with lumped model and ECM compatibility.
  • Fixed parameter reverse-parsing.
​
Live pipeline loading, simpler simulation API, and current variable support

​
Live pipeline loading

Pipeline pages now load progressively — only the next pending element is fetched at a time, with live updates as elements complete. This makes working with long pipelines much more responsive.

​
Simpler simulation API

The simulation API has been simplified from a two-step process to a single step. UCP input parameters are also now supported directly in simulation protocols, making it easier to configure and run simulations.Find out more →

​
Current variable support in pipeline

Pipeline objectives now support current as a variable, expanding the set of quantities that can be used in fitting and optimization workflows.
Improvements
  • Data loading on optimization pages improved with proper cleanup on navigation.
  • Job compute time is now logged for performance monitoring.
  • Renamed “model” to “parameterized model” and “equations” to “model” for clarity.
Fixes
  • Fixed optimization v1 data conversion error that caused the optimization list to appear empty.
  • Fixed design optimization bugs.
  • Fixed pipeline multistart plots consistency.
Improvements
  • Reverse parser now handles design objectives and metrics.
  • DesignObjective updated to use a metric dictionary format.
  • Sensitivity analysis improvements: fixed second-order calculation and added log-transform support.
  • Initial guess distributions (Log, Uniform) added for simpler configuration.
  • Simulation objectives can now expose additional output variables.
  • Multiple statistics supported in validation reports.
  • Initial voltage can now be inferred from the measurement data when evaluating custom parameters.
  • Parameter dependencies made optional for simpler objective definitions.
Fixes
  • Fixed a validation issue with objective inputs.
  • Fixed symbol evaluation in MSMR models.
  • Fixed time_series and steps key validation.
​
Simpler data API, faster large-result loading, and Sobol sensitivity analysis

​
Simpler data API

The data handling API has been simplified with improved validation, making it easier to upload and work with experimental data. Custom fields are now supported via an additional data column, so you can attach arbitrary metadata to your measurements.Find out more →

​
Faster loading for large results

Large results (such as simulation solutions and pipeline results) now load faster, with improved page load times and reduced memory pressure when working with large datasets.

​
Sobol sensitivity analysis

The pipeline now supports Sobol-based sensitivity analysis, enabling global sensitivity studies that quantify how each parameter contributes to output variance — including second-order interaction effects.
Improvements
  • Pipeline elements now use a more memory-efficient internal representation, reducing memory usage.
  • Optimization v2 refactored for a more consistent workflow.
Fixes
  • Fixed a bug where initial conditions that violate events caused errors.
  • Fixed user permissions at the organization level.
  • Fixed a frontend race condition when navigating between projects.
Improvements
  • ECM and lumped models are now compatible with the pipeline.
  • Interpolant numerical stability improved using smooth-step functions.
  • Optimiser architecture refactored for improved performance and scalability.
Fixes
  • Fixed entry/create serialization issues.
  • Fixed experiment parsing.
  • Fixed a bug with non-parametric inputs to the MSMR objective.
  • Improved interpolant error messages and numerical stability.
← Newer updatesOlder updates →