Keep multi-step simulation workflows moving with more control over parallel execution and failure recovery. Rescale Workflows now supports concurrent node limits, node-level failure controls, and failed-run resubmission, helping teams running design of experiments (DOEs) and other simulation campaigns manage how much work runs at once, contain failures at the step where they occur, and recover with greater precision.
Get started with reusable simulation pipelines using pre-built Workflow Templates from Rescale. Browse curated templates for parallel processing, dynamic hardware routing, centralized job configuration, and agent-assisted validation, then run them with your inputs or duplicate and adapt them for your team, turning common workflow patterns into working pipelines in minutes.
Simplify surrogate model development with AI Physics Auto Tune, coming soon to Rescale. Set a search duration and let AutoTune explore hyperparameter configurations in parallel, remove weak candidates early, fully train the strongest options, and select the best model across four production-ready architectures, without requiring deep AI/ML expertise.
Develop aerodynamic AI Physics models with HiLiftAeroML, an open-source dataset now available natively within Rescale AI Physics. Built from surface-based simulation data across four geometries and multiple angles of attack, it gives engineers a practical starting point for training AI surrogate models without first assembling and building their own dataset.
Connect CFD insight to downstream engineering decisions with an agentic Ansys CFX and Mechanical workflow on Rescale. The workflow combines the power of agentic engineering with deterministic simulation across multiple disciplines, analyzing a completed CFX design study, identifying key parameters and data gaps, kicking off follow-on studies, and then brings CFX pressure loads into Mechanical so hydraulic and structural results can be evaluated together.
Rescale has now extended its job submission environment to support multi-node MPI and Kubernetes-based workloads running on CoreWeave Cloud, bringing enterprise-grade HPC capabilities to a cloud infrastructure option that offers meaningfully lower cost for large-scale simulation. The capability has been validated end-to-end across multi-node configurations, with stability confirmed through scale testing and initial production use with enterprise customers.
Make simulation expertise easier to access across your organization. Rescale’s Simulation Guide Agent uses trusted technical documentation, user guides, templates, and team best practices to help users ask better questions, interpret results, select an appropriate workflow, configure jobs correctly, and verify recommendations before taking action, all within the existing simulation workflow on Rescale.
Build digital twin workflows that connect simulation and machine learning in one engineering process. Rescale helps teams move beyond isolated analyses by linking physics-based simulation, custom code, and data-driven models so digital twin initiatives can become more practical, repeatable, and useful in day-to-day engineering work.
Run multi-stage computational chemistry workflows without manual handoffs between steps. Rescale helps research teams connect simulation stages, compute environments, and downstream analysis so complex studies move forward in a more repeatable, scalable workflow instead of depending on ad hoc scripts and one-off operational effort.
Shorten the loop between design changes and aerodynamic insight. Rescale can connect updated CAD inputs, CFD workflows, and engineering data pipelines so new designs automatically trigger analysis, feed results into a shared data foundation, and return faster feedback to the teams shaping the product.
Explore more design options faster with local AI Physics inference. Rescale’s inference tooling lets engineers connect geometry, surrogate models, and outputs in a lightweight design workflow so they can run parametric studies and optimization loops directly on their own device (e.g. local workstation) without needing inference servers or HPC clusters.
Move files across multi-step workflows without manual hand-offs or separate storage setup. Rescale Workflows can now use cloud storage to pass files directly between job and workstation steps, helping engineering teams keep preprocessing, solve, and post-processing stages connected inside one reusable workflow.
Accelerate computational chemistry workflows with agentic support for molecular dynamics simulation on Rescale. Agents help researchers analyze results, propose next steps, configure new simulations, and critique outcomes with scientific context preserved, connecting HPC execution, scientific analysis, historical knowledge, and job setup into one guided workflow that keeps the researcher in control.
Build surrogate models for crash and structural simulation with Rescale’s integrated AI Physics workflow for transient FEA data. The updated crash use case brings GeoTransolver support for transient behavior, solver-specific metadata extraction, and training-ready formatting into one reusable pipeline—helping engineers move toward faster crash insight with less manual setup.
Deploy engineering agents with confidence with Rescale’s agent evaluation approach, anchored by the Rescale Agent Fidelity Toolkit (RAFT). Rescale RAFT combines response scoring, tool-use validation, ground-truth checks, and continuous monitoring to systematically verify that agents deliver relevant answers and take correct actions, providing the quality discipline that production agentic engineering demands.
Move simulation data between your desktop and the cloud faster with Rescale Interlink, an open-source hybrid CLI and GUI tool for managing files and jobs on Rescale. FIPS 140-3 compliant and available on Windows, Mac, and Linux, Interlink delivers multithreaded parallel transfers, a visual file browser, auto-download of completed job results, and full job submission, all in one lightweight, community-driven application.
Resolve simulation failures in minutes, not hours. Rescale’s Job Troubleshooting Agent analyzes solver logs, pinpoints the failure, explains what went wrong in plain language, and recommends corrective actions, all within the Rescale Assistant. Engineers review and approve fixes before resubmission, keeping human judgment in the loop while eliminating the manual overhead of failure diagnosis.
Detect diverging and stalled simulations in real time, before they waste hours of compute. The Simulation Monitor automation runs alongside iterative solver jobs on Rescale, providing a live dashboard with residual plots, tabular data switching, and automated CSV exports without requiring an interactive workstation. It supports Ansys Fluent, STAR-CCM+, and CFX today, with a plugin-based framework designed for fast extension to additional solvers.
Start training surrogate models faster with open-source AI Datasets now available in Rescale AI Physics. Engineers can now train models directly from the DrivAerML dataset, a subset of the high-fidelity open-source public dataset for automotive aerodynamics based on 500 parametrically morphed variants. Providing a great way to get started with AI Physics without having to first generate your own simulation dataset.
Power complex and high-accuracy AI surrogate models for CFD and FEA with GeoTransolver model architecture. Now available in Rescale AI Physics as part of the NVIDIA PhysicsNeMo library, GeoTransolver is a geometry-aware transformer architecture that uses GALE attention to model relationships across complex 3D geometries and unstructured meshes. It is designed to generalize across changing operating conditions and learn transient, geometry-aware physical behavior, making it well-suited for surrogate models that need to predict aerodynamic, structural, or deformation behavior across a range of design variants.