Skip to main content

Release - March 6, 2026

LTX Video 2.3, batch AI prompt generation, grouping nodes, live workflow progress tracking, and more.

Updated this week

What's new?

  • New video models: LTX Video 2.3 & 2.3 Fast. You can now generate text-to-video and image-to-video content using LTX Video 2.3 (enhanced visuals, cleaner audio, portrait support) and LTX Video 2.3 Fast (speed-optimized, up to 20 seconds duration), giving you more flexibility for high-quality video production.

  • Live workflow progress tracking: You can now monitor workflow executions in real time directly alongside your tool calls, with step-by-step status updates, cost monitoring, and interactive output previews, so you always know what's running and what it's costing.

  • Group image behavior controls: You can now configure whether images in a group are sent as variants or references in the Image Prompt Edit node, giving you more precise control over creative outputs.

  • AI-powered prompt generation: You can now use the new Prompt Generator node in blueprints to create many prompts at once with a single instruction prompt so you save a ton of time generated relevant prompts with less manual iteration.

Bug fixes & improvements

  • Composition canvas fixes: Resolved issues with rectangle positioning, text overflow, and the canvas disappearing after a node runs.

  • Prompt image paste restored: Fixed an issue preventing images from being pasted into the prompt field.

  • Reduced stuck inference runs: Improved reliability of inference run completion, reducing instances that required manual intervention.

  • Crash fix on legacy nodes: Resolved a page crash that occurred when selecting older prompt edit nodes.

  • Incident banner readability: The incident banner now displays the actual incident title and is easier to read at a glance.

What's next?

  • Workspace creative unit transfers: You can now split and allocate creative units across workspaces, with support for auto-transfers from invoices and manual allocation management, making subscription management more flexible for larger teams.

  • Browse and inspect models via MCP – Agents can now search and filter available models by modality, type, and capabilities directly through MCP, making it easier to find the right model for your workflow.

Did this answer your question?