> For clean Markdown of any page, append .md to the page URL. > For a complete documentation index, see https://docs.vellum.ai/llms.txt. > For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.vellum.ai/_mcp/server. # Changelog | April, 2025 ## Dynamic Model Select on Inline Prompt Nodes **April 29th, 2025** Inline Prompt Nodes now accept expression inputs to represent the Model invoked in the Prompt. Previously, you could only statically choose a model from a list of available models. Now, you can also dynamically reference a model from an upstream Node or Workflow Input. ## Ports on Workflow Nodes **April 28th, 2025** All Workflow Nodes now support Ports. Previously you had to define a Conditional Node for control flow in a workflow. Now you can define Ports using Expression Inputs to define the control flow you want. ## Non-Streaming Prompt Nodes **April 28th, 2025** Prompt Nodes now support the option to run as streaming or non-streaming. The latter is considered more performant during parallelization or Workflows that feature Map Nodes. ![Non-Streaming Prompt Nodes](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2025-04/non-streaming-prompt-nodes.png) ## Test Suite Run Progress **April 25th, 2025** It's now possible to see the progress of an evaluation report run while it's actively running. Hovering over the progress indicator, a tooltip will display the number of test cases being run and the number of test cases completed. ![Test Suite Run Progress](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2025-04/test-suite-run-progress.png) ## Run From Node Re-enabled for SDK-Enabled Workflows **April 22nd, 2025** Back in March, we needed to temporarily disable our Run From Node feature while we reimagined how it should be designed in this new paradigm. Today, we are excited to announce that we are bringing *back* this beloved feature for all SDK-Enabled Workflows. Just as before, to invoke the a Workflow starting from a given Node, click the mini play button that appears on the Node when a previous execution is loaded. This will use the "State" we saved from the previous execution in order to run the Workflow from that exact point in time going forward. ![SDK Run from Node](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2025-04/sdk-run-from-node.png) ## o4 Mini and o3 Models on Vellum **April 16th, 2025** We've added [OpenAI's newest Models](https://openai.com/index/introducing-o3-and-o4-mini/) to Vellum: * o3 * o4 Mini ## Metric Mapper Revamp **April 15th, 2025** We've introduced a revamped UI that makes it easier to configure Metrics within the context of Evaluations. A Metric's inputs are now automatically mapped to existing variables based on their name and type. We also now clearly indicate which inputs are required and simplify the process of using constant values. All in all, we hope that with these changes, it requires fewer clicks to set up a Metric and understand where it's getting its inputs from! ![Revamped Metric Mapper](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2024-04/metric-mapper.png) ## Llama 4 Maverick via Groq on Vellum **April 14th, 2025** We've added support for Llama 4 Maverick 17B Instruct via Groq to Vellum ## GPT 4.1 Models on Vellum **April 14th, 2025** We've added the new GPT 4.1 Models via OpenAI to Vellum: * GPT-4.1 * GPT-4.1 (2025-04-14) Snapshot * GPT-4.1 Mini * GPT-4.1 Mini (2025-04-14) Snapshot * GPT-4.1 Nano * GPT-4.1 Nano (2025-04-14) Snapshot ## xAI Models on Vellum **April 14th, 2025** We've added the new xAI models to Vellum: * Grok 3 Beta * Grok 3 Fast Beta * Grok 3 Mini Beta * Grok 3 Mini Fast Beta ## Test Case Error Filtering for Evaluation Reports **April 10th, 2025** We've introduced a new feature that allows you to filter and select failed test cases in evaluation reports. This enhancement enables you to re-run only the errored test cases, which is particularly beneficial when dealing with provider rate limits on extensive evaluation reports. ![Evaluation Report Error Filtering](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2024-04/error-filter.png) ## Sandbox Cost Tracking for Workflows **April 8th, 2025** We've now added cost tracking support for your Subworkflow and Deployment Subworkflow Nodes from inside our Workflow Sandbox. This is an aggregate sum of total costs for your workflow invoked by your Subworkflow Nodes, available immediately after running your workflow! ![Sandbox Workflow Cost](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2024-04/sandbox-workflow-costs.png) ## Llama 4 Models on Vellum **April 5th, 2025** We've added the new Llama 4 models to Vellum. These models are the latest and greatest from Meta. The new models are available from the following providers: * Fireworks AI * Llama 4 Scout * Llama 4 Maverick * Groq * Llama 4 Scout ## Add Node on Edge Drop **April 3rd, 2025** Previously, the only way to add a Node was to drag and drop a new Node from the Nodes Panel. We've now introduced a faster way to add and connect Nodes: simply drag an edge from an existing Node and drop it where you want the new Node to be. A Node selection menu will appear at that location, allowing you to choose the type of Node you want to add. The new Node will be automatically placed and connected to the edge. ## Protected Release Tags **April 1st, 2025** You can now specify Release Tags that should be considered "protected." This feature is meant to be used in conjunction with our recently released [Deployment Release Reviews](/changelog/2025/2025-03#deployment-release-reviews) feature. ![Protected Release Tags Setting](https://storage.googleapis.com/vellum-public/help-docs/changelogs/2025-04/protected-release-tags-setting.png) A protected Release Tag cannot be assigned to a Prompt or Workflow Deployment Release unless that Release has at least one approval from a Reviewer and no outstanding change requests. This feature is useful for ensuring that critical or sensitive Deployments are reviewed and approved before going live. ## Grouping a Selection into a New Subworkflow Node **April 1st, 2025** Previously, creating Subworkflows and organizing Nodes within them required manual copy-pasting of Nodes. We've now introduced a powerful new way to organize your Workflows by grouping Nodes into Subworkflows. You can now select multiple Nodes and automatically convert them into a new Subworkflow Node, which will maintain all the existing connections. Additionally, you can drag and drop selected Nodes onto existing Subworkflow Nodes to move them inside. This feature helps you better organize complex Workflows into logical components while reducing visual clutter in your Workflow canvas. When grouping Nodes, all existing connections and logic are automatically maintained, and you can easily modify Subworkflow contents through simple drag and drop interactions.