Luma AI Dream Machine
last edited
Luma AI Dream Machine was the original name of Luma's hosted video generation product. Luma's own LLM info page now lists the newest Ray model as the current video model and calls the older product deprecated. This page covers the current Luma lineup, its funding and compute plans, and its assistant connections. Related open-source tooling is covered in AnimateDiff.
Current model
Luma released the Ray3.2 video model in June 2026. It succeeds the Ray3.14 model, which Luma launched in January. That earlier model added native 1080p output and ran faster. It also cost less to use.
The newer model supports up to 16 keyframes. It adds frame-level control, HDR with 16-bit EXR export, and Modify and Reframe tools. A public API ships with the model. Luma states that earlier releases, including Dream Machine, are deprecated. Details are on the Luma Ray model page and the Luma LLM info page.
Funding and compute
Luma announced a $900M Series C in November 2025. The round came with a partnership with Humain on Project Halo, a 2GW compute supercluster. Per Luma's post, deployment begins in early 2026. It finishes between 2028 and 2029. Details are in the Luma Series C announcement.
Assistant integrations
Luma announced a connection to ChatGPT and Claude through MCP. The Luma MCP integration post describes it. The link lets those assistants call Luma's video tools.
Hosted generation on Luma can sit beside open tooling for local motion experiments. Workflows are covered in AI video production pipelines and the AI Video Generation reference. Production teams often compare Luma with other proprietary vendors, as covered in the Production Reliability Comparison and on the leading proprietary models page.