- Capabilities
- Getting started
- Architecture center
- Platform updates
Announcements
REMINDER: Sign up for the Foundry Newsletter to receive a summary of new products, features, and improvements across the platform directly to your inbox. For more information on how to subscribe, see the Foundry Newsletter and Product Feedback channels announcement.
Share your thoughts about these announcements in our Developer Community Forum ↗.
Additional open-weight models now available in AIP through Fireworks
Date published: 2026-09-10
Select models from Z.ai and Moonshot AI are now available in AIP through Fireworks on non-georestricted enrollments.
Model overviews
GLM-5.3 ↗ is an advanced open-weight frontier reasoning model released by Z.ai that delivers state-of-the-art performance for complex software engineering, long-horizon autonomous agents, and cybersecurity tasks.
GLM-5.3 Flash ↗ is an open-weight, natively multimodal AI model released by Z.ai with strong performance in coding and agent tasks at a reduced cost and higher speed.
Kimi K3 ↗ is an open-weight multimodal AI model released by Moonshot AI. It is designed for complex reasoning, long-horizon coding, and agentic knowledge work.
Getting started
To use these models:
- Confirm that your enrollment administrator has enabled the relevant model family.
- Review token costs and pricing.
- See the complete list of models available in AIP.
Your feedback matters
We want to hear about your experiences using language models in the Palantir platform and welcome your feedback. Share your thoughts with Palantir Support channels or on our Developer Community ↗ using the language-model-service tag ↗.
Gemini 3.8 Flash from Vertex AI is now available in AIP
Date published: 2026-09-10
Gemini 3.8 Flash is now available in AIP for non-georestricted, US georestricted, EU georestricted, IL2, IL4, and IL5 enrollments with Google Vertex AI enabled.
Model overview
Gemini 3.8 Flash is the next iteration in the Gemini 3 model family, with improvements in reasoning and coding. For more information, review Google's Gemini 3.8 Flash model card ↗ and Google's model announcement ↗.
Availability
Gemini 3.8 Flash is available on:
- Non-georestricted commercial enrollments with Google Vertex AI enabled
- US georestricted commercial enrollments with Google Vertex AI enabled
- EU georestricted commercial enrollments with Google Vertex AI enabled
- IL2, IL4, and IL5 enrollments with Gemini through Google Vertex AI enabled
Getting started
To use this model:
- Confirm that your enrollment administrator has enabled the relevant model family.
- Review token costs and pricing.
- See the complete list of models available in AIP.
Your feedback matters
We want to hear about your experiences using language models in the Palantir platform and welcome your feedback. Share your thoughts with Palantir Support channels or on our Developer Community ↗ using the language-model-service tag ↗.
GPT-5.6 series now available in AIP on IL2, IL4, and IL5 enrollments
Date published: 2026-09-10
GPT-5.6 Sol, GPT-5.6 Terra, and GPT-5.6 Luna are now available in AIP on IL2, IL4, and IL5 enrollments with Azure OpenAI enabled.
Model overviews
The GPT-5.6 series is OpenAI’s newest family of models. It includes a frontier model (Sol) for advanced workloads, a balanced model (Terra) for intelligence and cost, and a cost-effective model (Luna) for high-volume use cases.
For more information, review OpenAI’s model documentation ↗ and OpenAI’s GPT-5.6 announcement ↗.
Getting started
To use these models:
- Confirm that your enrollment administrator has enabled the relevant model family.
- Review token costs and pricing.
- See the complete list of models available in AIP.
Your feedback matters
We want to hear about your experiences using language models in the Palantir platform and welcome your feedback. Share your thoughts with Palantir Support channels or on our Developer Community ↗ using the language-model-service tag ↗.
Validate production LLM workflows with Pipeline Builder evaluation suites
Date published: 2026-09-10
Large language model (LLM) evaluation suites are now generally available in Pipeline Builder. Use evaluation suites to test the outputs of Use LLM nodes before deploying changes to your pipeline Each run writes detailed results to a Foundry dataset, helping you identify opportunities to improve your LLM workflows.
Test LLM workflows in an isolated environment
Evaluate changes without affecting your production pipeline. Import an existing Foundry dataset as testing data or enter test data manually, then select a dataset for the evaluation results.
Add testing data, an output dataset, and evaluators for a Use LLM node.
Compare outputs with built-in evaluators
Add one or more evaluators to compare generated outputs with expected results or other passing conditions. For example, use Exact string match for direct comparisons or LLM-as-a-judge to determine whether a user-defined condition is satisfied.
Configure evaluators in evaluation suites.
Analyze results in Foundry
Preview results directly in Pipeline Builder or use the output dataset in other Foundry applications, such as Contour. Analyze performance and trends, investigate individual results, and identify changes that can improve your workflows.
Learn more about LLM evaluation suites in Pipeline Builder.
Share your feedback
As we continue to add features to Pipeline Builder, we want to hear about your experiences and welcome your feedback. Share your thoughts with Palantir Support channels or our Developer Community ↗ using the pipeline-builder ↗ tag.
Browse, compare, and share committed code with the code viewer in Code Workspaces
Date published: 2026-09-10
The new code viewer is now available in Code Workspaces, providing a read-only interface for browsing, comparing, and sharing committed code from repositories stored in Foundry, whether the code was developed in Foundry or locally. You can start browsing code immediately while a VS Code workspace loads in the background.
Browse committed code
Open a repository in Code Workspaces, then select the Code viewer tab in the page header. From the code viewer, you can:
- Browse files on a branch, tag, or commit
- Search by file name or across file contents
- View supported files, including Markdown files and Jupyter® notebooks, as rendered documents or source code
- Preview published functions in function repositories
Editor changes appear in the code viewer after they are committed and synchronized with the repository.
Browse files and view committed code in the code viewer.
Compare repository history
You can also use the code viewer to explore changes between two points in repository history. Select Compare changes, choose the references you want to compare, and select a changed file to view a side-by-side diff.
Compare two repository references with a side-by-side diff.
Share stable links to code
Create a permalink to an exact commit and line range so others can reference the same code, even as the branch advances. Select a line number, or hold Shift while selecting another line to choose a range, then select Line actions > Copy permalink.
Copy a permalink to selected lines.
Browse while VS Code starts
By default, the code viewer opens when you open a stopped VS Code workspace, allowing you to browse committed code while the workspace session starts in the background. The VS Code tab icon displays the workspace status. When the workspace is ready, select VS Code to begin editing.
To disable this behavior, select Settings > Preferences, then disable the setting to Open code viewer while workspace is loading.
The code viewer is available alongside VS Code, JupyterLab®, and RStudio® Workbench. Automatic background startup applies only to VS Code workspaces.
Learn more about the code viewer in Code Workspaces.
Share your thoughts
We want to hear about your experience using the code viewer in Code Workspaces. Share your feedback with Palantir Support channels or on our Developer Community ↗ using the code-workspaces ↗ tag.
Detect sensitive data using language model match conditions in Sensitive Data Scanner
Date published: 2026-09-10
To detect sensitive data that is difficult to model as a regular expression or fixed list of values, you can now use a natural language prompt and set examples for large language models to evaluate datasets, virtual tables, and media sets using match conditions in Sensitive Data Scanner. Generally available across Foundry enrollments the week of September 7, language model match conditions use a large language model and a natural language prompt to classify whether scanned content, column names, or both contain sensitive data. When scanning media sets, Sensitive Data Scanner supports selecting a vision-capable model to classify image content directly or using a text-only model to classify extracted text from documents.
Get started with language model match conditions
To create a language model match condition, choose Language model as the condition type, select a model, and define the matching logic. You can configure the condition to evaluate:
- Content, meaning the data values within a column
- Column names
- Content and column names, where both must match
- Content or column names, where either can match
Additionally, Sensitive Data Scanner enables you to define examples and counter examples to clarify the information the model should detect.
Configure the model and prompt for your language model match condition in Sensitive Data Scanner.
Review the create match conditions documentation to learn more about configuring language model match conditions.
Share your thoughts
As we continue to add features to Sensitive Data Scanner, we want to hear about your experiences and welcome your feedback. Share your thoughts with Palantir Support channels or our Developer Community ↗ using the sensitive-data-scanner ↗ tag.
Introducing AIP Evolve: Coordinate AI FDE agents to improve AI systems in Foundry
Date published: 2026-09-08
AIP Evolve is now generally available for enrollments with AIP enabled and access to AI FDE. AIP Evolve coordinates fleets of AI FDE agents to improve AI systems in AIP. Define a target, optimization goal, validation strategy, and operational constraints, then review the resulting proposal and agent activity before merging changes.
Early adopters have used AIP Evolve to autonomously cut AI costs, improve eval performance, and migrate workloads to open source models. Learn more about AIP Evolve.
Key features
AIP Evolve supports iterative improvement workflows with the following features:
- Guided setup for model migration, cost reduction, latency reduction, evaluation score improvement, and custom goals.
- Flexible validation using selected test cases or existing evaluation suites.
- Configurable scoring criteria and acceptable output divergence.
- Agent constraints that control permitted change types and the maximum number of iterations.
- A proposal view containing proposed changes, validation results, output comparisons, supporting evidence, and confidence assessments.
- An interactive agent graph for monitoring progress and inspecting agent goals, insights, and artifacts.
- Integration with Global Branching for reviewing proposed changes before merging.
- The ability to resume an evolution in AI FDE with additional instructions.
Requirements
To use AIP Evolve:
- Enable AIP and ensure that you can access AI FDE.
- Install the AIP Evolve Marketplace product in an Ontology.
- Ensure that you have access to the target resource and any data or evaluation suites used for validation.
Getting started
The AIP Evolve setup screen, showing the Review stage with a fully specified evolution.
Open AIP Evolve and select New to create an evolution. Select the Foundry resource you want to evolve, then configure the following:
- Goal: Choose a predefined optimization goal or describe a custom objective.
- Validation strategy: Define the test data, scoring approach, and acceptable output divergence.
- Agent constraints: Select the types of changes agents may propose and set an iteration policy.
Review the generated prompt, then select Evolve. AIP Evolve opens AI FDE in a new tab and starts the evolution. You can also select Write custom prompt to provide your own instructions.
An example AIP Evolve proposal, where the system presents a model swap for cost reduction and the evidence that supports why the change is safe to make.
Open Evolutions to monitor active and completed evolutions. Use Proposal to review results and proposed changes, or Agent graph to inspect the agents involved in the evolution. When a proposal is ready, open it in Global Branching for final review and merging.
The AIP Evolve agent graph, showing each of the subagents that were spawned along the way to optimize the target AI component.
Your feedback matters
As we continue developing AIP Evolve, we welcome feedback about your experience. Share your thoughts with Palantir Support channels or our Developer Community ↗ using the aip-evolve tag ↗.
Introducing Vulcan, a CAD review application for the Ontology
Date published: 2026-09-08
Vulcan is a 3D visualization application that renders engineering geometry directly from Ontology data. Mesh models (GLB, GLTF, STL, OBJ, PLY, and 3MF), 2D engineering drawings (DXF), and point clouds (LAS and LAZ) can be loaded without conversion, and a single scene can contain all the models needed for a workflow. Vulcan is available in beta across all enrollments starting the week of September 7.
When saved, annotations, measurements, and section planes created in Vulcan are written to the Ontology as objects. A flagged defect can be used to trigger an alert, a critical dimension can be tracked in a dashboard over time, and a saved cross-section can be retrieved or shared with another team. Vulcan runs as a standalone application and can also be embedded in Workshop as a widget, where camera position, part selection, and measurement results can be connected bidirectionally to Workshop variables.
The part tree lists each part instance in the assembly; each instance corresponds to an object in the Ontology. Model source: Printables. Released under a Creative Commons public-domain license.
Vulcan has Ontology-backed tools to support cross-team workflows:
- Measure distances on models
- Create annotations
- Slice models with section planes
- Apply colors to parts with Ontology actions
Annotations and measurements on a CAD model in Vulcan. Model source: Printables. Released under a Creative Commons public-domain license.
See how Vulcan can support your production workflows:
- Markup models to provide instructions for assembly or repair procedures
- Share annotations and measurements across design-review teams
- Color models according to inventory levels or failure rates
- Visualize the results of automated design-verification runs in the Palantir platform
- Use components and models imported into Vulcan for further analysis across the Foundry application suite
To get started, load a CAD model into the viewport and select Add to scene. To learn more about the tools, the Workshop widget, and the Ontology object types and actions that Vulcan reads and writes, review the Vulcan documentation.
Your feedback matters
We want to hear about your experiences using Vulcan in the Palantir platform and welcome your feedback. Share your thoughts with Palantir Support channels or on our Developer Community ↗ using the vulcan tag ↗.
Visual rebasing makes Workshop changes easier to resolve
Date published: 2026-09-03
When multiple builders update the same Workshop module on different branches, rebasing incorporates changes from main into a branch before that branch is merged. Workshop now provides visual rebasing to help reconcile granular changes to widgets, sections, and variables between main and your branch.
Previously, you had to inspect the module’s JSON definition in the Changelog panel to determine what changed. Visual rebasing brings this workflow into the module editor, where you can accept, reject, or combine changes while seeing how they affect your module.
The rebase dialog lets builders start a visual rebase or reject all changes from main.
Understand changes in context
During rebasing, change icons appear beside the affected settings in widget and section configuration panels. The icons distinguish additions, modifications, deletions, shifts, and conflicts, so you can locate changes without leaving the configuration you are reviewing. Select an icon to open a detailed view of main and your branch side by side.
- Conflict: The same property for the same component was edited on
mainand the branch. - Addition: A property or component was added.
- Modification: A property or component was edited.
- Deletion: A property or component was deleted.
- Shift: A section or widget was moved to a different parent.
- Branch: A property or component was resolved using the branch version.
The change icon legend identifies differences between the two versions.
The review changes dialog compares the metric card configuration on main with the configuration on your branch.
Variable conflicts use the same workflow. Select a conflicting variable to open its editor and compare the complete definition and settings for each branch.
How to preview, accept, and combine changes
By default, Workshop automatically accepts non-conflicting changes from main and merges them into your branch. You only need to take action when the same configuration field, variable definition, or layout position was changed on both branches.
Select Main branch or Your branch to preview a configuration temporarily and see its effect on the module. You can accept either version as-is or use one as a starting point for further edits. Those edits create a Current session state that can combine changes from both branches.
To replace the entire module configuration from main with the branch configuration, select Reject all changes from main when you start the rebase. This causes the branch configuration to override changes from main when you merge the branch. Use this option only when you want to discard all incoming changes.
What's next
Future updates to visual rebasing will support layout shifts and changes to module-level settings.