Luma Ray3.2
Luma's current video model, released June 2026. Generates 5 or 10 second clips at up to 1080p from text or images, with six aspect ratios, native HDR in 16-bit colour, EXR export in ACES2065-1 for grading and VFX, looping, and multi-keyframe direction.
- Up to
- 10 seconds
- Resolution
- 1080p
- Aspect ratios
- 9:16, 3:4, 1:1, 4:3
- Weights
- Not released
- API
- Yes
- Commercial use
- Permitted subject to conditions
Model ID ray-3.2 on the Luma API.
Commercial use rights included on individual plans.
- Depends on your plan tier. Commercial use rights included on individual plans
- Output may not be used to train AI. Provisioned throughput includes a no-training guarantee
Specification
- What it takes in
- text, image, video
- What it makes
- video
- Maximum duration
- 10 seconds
- Maximum resolution
- 1080p
- Aspect ratios
- 9:16, 3:4, 1:1, 4:3, 16:9, 21:9
- Controls
- image-to-video, keyframes, extend, reference-character-consistency
- Weights
- Not released
- API
- Yes
- Pricing
- credit-based subscription for the web app (Plus $25/mo yearly for 10,000 credits, Pro $75, Ultra $250); API pay-as-you-go per generation - 5s from $0.06 (360p) to $1.20 (1080p), 10s from $0.18 to $3.60, HDR 5s $0.60-$2.40. Provisioned throughput available from $2,100-$3,800 per unit per month, minimum 8 units
- Released or updated
- 2026-06
- Prompt limit
- 6,000 charactersVendor figure, checked 2026-08-16Collected from the vendor's own documentation on 16 August 2026. Vendors change these between model versions without announcing it, so check the vendor's current API reference before you rely on it. Check the vendor's documentation
- Prompt length the vendor recommends
- 40 to 100 wordsThe recommended length is well below the cap on every model that publishes both. The cap is the point of rejection, not the target: writing to it produces worse output, not more detail.
Strengths
- Native HDR generation in 16-bit colour and EXR export at ACES2065-1, unusual among video models
- Up to 64 keyframes with explicit frame indexes for precise timing control
- Six aspect ratios including 21:9 and 1:1
- Loop generation
- Forward and backward extension by passing a prior generation ID as a keyframe
- 360p draft tier at $0.06 per 5s clip makes iteration very cheap
Weaknesses
- No audio generation documented
- 10-second cap, and 10s is incompatible with HDR, start_frame and end_frame
- HDR is restricted to 5s at 720p/1080p and rejects 360p/540p and looping
- 1080p 10s clips cost $3.60 each, expensive relative to competitors
- Luma's own pages disagree on keyframe limits and Modify Video length
Best for
- VFX and grading pipelines needing HDR/EXR
- precise keyframe-timed motion
- looping assets
- cinematic shots at multiple ratios
Where the sources disagree
Recorded rather than resolved. Picking the tidier number would hide the fact that the vendor's own pages do not agree.
- Luma's llm-info page states Multi-Keyframe supports "up to 16 keyframes" while the Luma API video generation guide documents 1-64 keyframes - https://lumalabs.ai/llm-info vs https://docs.agents.lumalabs.ai/guides/videos/generation/index.md
- Luma's llm-info page says Ray3.2 is the current model and that Ray3.14 was superseded, but lumalabs.ai/pricing still lists Ray3.14 as a selectable model at 80 credits/sec at 1080p and Luma's own field guide calls Ray3.14 "your default workhorse for video generation" - https://lumalabs.ai/llm-info vs https://lumalabs.ai/learning-center/articles/luma-video-models-field-guide
What to understand before using it
Explainers naming the mechanisms this model's own documentation mentions.
MCP servers that front Luma AI
Servers in the media MCP directory whose own description or homepage names this vendor. Absence of a server here is not evidence that none exists: most publish no tool list at all.
-
Luma
MCP server for Luma Dream Machine AI video generation.
-
Picsart GenAI
Generate and edit images, videos and audio with 150+ AI models.
-
Shortgenius
Generate AI short-form videos and auto-publish them to TikTok, YouTube, Instagram, Facebook and X.
-
Varosity
One key for AI video, voice, music, and image across 40+ frontier models. BYOK zero markup.
-
Vivideo
AI video generation via the Vivideo API: auto or manual mode, avatars, voices, brand kits.
See it from the vendor
Shotcraft hosts no sample output. Judging a model means watching its own showreel with your own eyes, and a clip re-hosted here would be a clip we chose.
Nearby in the directory
Luma Dream Machine / Ray2
Luma's original consumer video product (Dream Machine) and the Ray2 model family behind it, with API model IDs ray-2 and ray-flash-2 and resolution op
Video generation
Luma Ray3.14
Ray3 line model that preceded Ray3.2. Luma's field guide describes it as the default workhorse with 5-10 second text-to-video, 5 second image-to-video
Video generation
Adobe Firefly Video Model
The Adobe Firefly Video Model generates video from a text prompt, optionally guided by an image used as the first or final frame. Adobe's Firefly Serv
Video generation