Skip to content
Model infoModelsSnapshot: 15 Oct 2026

Luma AI / Ray

Ray 3.2

Video generation, keyframe guidance, editing, extension and aspect-ratio reframing.

Video generation
availableClosed weightsVideo generationVideo editing
Official documentation (opens in a new tab)Reviewed 12 Oct 2026
10 of 23 specification fields collected
Output resolution360p (draft), 540p, 720p, 1080pSee specification notes below
Clip duration5 or 10 seconds for generation; start/end anchors and HDR require 5 secondsSee specification notes below
Input billingSee ratesLuma Agents API
Output billingSee ratesPublished units and conditions below

Specifications

Provider-documented facts. “Not collected” means we have not verified that field yet, not that the feature is unsupported.

Context / input limit
Not collected
Maximum output
Not collected
Input modalities
Text, Image, VideoSource for Input modalities (opens in a new tab)
Output modalities
VideoSource for Output modalities (opens in a new tab)
Reasoning
Not collected
Tool calling
Not collected

Agent orchestration, hosted GLM tools and built-in web grounding do not establish arbitrary function calling by this model.

Structured output
Not collected

A structured API response envelope is not schema-constrained model text output.

Streaming
Not collected

Polling generation status does not establish progressive output streaming.

Batch processing
Not collected

Asynchronous jobs and concurrent requests do not establish a dedicated Batch API.

Fine-tuning
Not collected
Knowledge cutoff
Not collected
Release date
2026-06-09Source for Release date (opens in a new tab)
Parameters
Not collected
Architecture
Not collected
Languages
Not collected
Context caching
Not collected

Reusable uploaded media and expiring download links do not establish model-inference caching or discounted cache rates.

Output formats
MP4, EXRSource for Output formats (opens in a new tab)

EXR requires compatible HDR generation/editing. Reframe and extend are SDR-only.

Maximum prompt characters
6,000 charactersSource for Maximum prompt characters (opens in a new tab)
Output resolutions
360p (draft), 540p, 720p, 1080pSource for Output resolutions (opens in a new tab)
Video duration
5 or 10 seconds for generation; start/end anchors and HDR require 5 secondsSource for Video duration (opens in a new tab)
Video input duration
Up to 18 seconds for uploaded edits; 10 seconds for reframingSource for Video input duration (opens in a new tab)

Reframe is separately limited to 10 seconds; the marketing site advertises V2V up to 20 seconds.

Reference images
Up to 64 keyframe images via paired frame indexesSource for Reference images (opens in a new tab)

The marketing page advertises 16; current API generation/editing guides explicitly allow 64. Start/end anchor fields and the multi-keyframe list are mutually exclusive.

Aspect ratios
9:16, 3:4, 1:1, 4:3, 16:9, 21:9Source for Aspect ratios (opens in a new tab)

Access & pricing

Prices belong to an endpoint and billing tier. A provider listing does not guarantee access for every account or region.

Luma Agents API

Provider source (opens in a new tab)
API model ID
ray-3.2
Access
Documented API access; an eligible account, API key and billing setup are required. No paid generation was performed.
Regions
Not collected
API endpoint URL
POST https://agents.lumalabs.ai/v1/generationsSource for API endpoint URL (opens in a new tab)
API operation
videoSource for API operation (opens in a new tab)
Asynchronous generation
SupportedSource for Asynchronous generation (opens in a new tab)
Download URL lifetime
1 hour; poll again to refreshSource for Download URL lifetime (opens in a new tab)
HDR output
SupportedSource for HDR output (opens in a new tab)

720p/1080p and 5 seconds only; EXR requires hdr=true.

Seamless looping
SupportedSource for Seamless looping (opens in a new tab)

Rejected with 10s, HDR, end_frame or multi-keyframe input.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Video generation360p (draft) SDR 5s1$0.065s video
Video generation360p (draft) SDR 10s1$0.1810s video
Video generation540p SDR 5s1$0.155s video
Video generation540p SDR 10s1$0.4510s video
Video generation720p SDR 5s1$0.305s video
Video generation720p SDR 10s1$0.9010s video
Video generation1080p SDR 5s1$1.205s video
Video generation1080p SDR 10s1$3.6010s video
Video generation720p HDR 5s2$0.605s video
Video generation1080p HDR 5s2$2.405s video
Video generation720p HDR + EXR 5s3$0.905s video
Video generation1080p HDR + EXR 5s3$3.605s video

Rate conditions

  1. video only. SDR. Duration-specific total; do not assume linear scaling between 5 and 10 seconds. Published video rates are subject to change ahead of general availability.
  2. video only. HDR. Duration-specific total; do not assume linear scaling between 5 and 10 seconds. Published video rates are subject to change ahead of general availability.
  3. video only. HDR + EXR. Duration-specific total; do not assume linear scaling between 5 and 10 seconds. Published video rates are subject to change ahead of general availability.
Pricing source (opens in a new tab)

Data & enterprise terms

API Terms section 10 commits not to train on covered API inputs or outputs. The pricing page separately labels no-train as a Provisioned Throughput benefit and says no guarantee for pay-as-you-go; this conflict requires contractual confirmation.

Read the service policy (opens in a new tab)

Luma Agents API

Provider source (opens in a new tab)
API model ID
ray-3.2
Access
Documented API access; an eligible account, API key and billing setup are required. No paid generation was performed.
Regions
Not collected
API endpoint URL
POST https://agents.lumalabs.ai/v1/generationsSource for API endpoint URL (opens in a new tab)
API operation
video_editSource for API operation (opens in a new tab)
Asynchronous generation
SupportedSource for Asynchronous generation (opens in a new tab)
Download URL lifetime
1 hour; poll again to refreshSource for Download URL lifetime (opens in a new tab)
Video input duration
At most 18 seconds for URL/base64 inputSource for Video input duration (opens in a new tab)
Input video limits
At most 200 MB for uploaded videoSource for Input video limits (opens in a new tab)
Video editing controls
pose, depth, normals, trajectory, faceSource for Video editing controls (opens in a new tab)
HDR output
SupportedSource for HDR output (opens in a new tab)

Guide restricts HDR edits to 720p/1080p. Lower-resolution HDR price rows conflict with validation and are not published as usable rates.

Aspect ratios
Derived from source; requested aspect_ratio is ignoredSource for Aspect ratios (opens in a new tab)

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Video editing360p (draft) SDR 5s1$0.545s video
Video editing360p (draft) SDR 10s1$1.0810s video
Video editing540p SDR 5s1$0.725s video
Video editing540p SDR 10s1$1.4410s video
Video editing720p SDR 5s1$1.085s video
Video editing720p SDR 10s1$2.1610s video
Video editing1080p SDR 5s1$2.165s video
Video editing1080p SDR 10s1$4.3210s video
Video editing720p HDR 5s2$2.165s video
Video editing720p HDR 10s2$4.3210s video
Video editing1080p HDR 5s2$4.325s video
Video editing1080p HDR 10s2$8.6410s video
Video editing720p HDR + EXR 5s3$3.245s video
Video editing720p HDR + EXR 10s3$6.4810s video
Video editing1080p HDR + EXR 5s3$6.485s video
Video editing1080p HDR + EXR 10s3$12.9610s video

Rate conditions

  1. video_edit only. SDR. Duration-specific total; do not assume linear scaling between 5 and 10 seconds. Published video rates are subject to change ahead of general availability.
  2. video_edit only. HDR. Duration-specific total; do not assume linear scaling between 5 and 10 seconds. Published video rates are subject to change ahead of general availability.
  3. video_edit only. HDR + EXR. Duration-specific total; do not assume linear scaling between 5 and 10 seconds. Published video rates are subject to change ahead of general availability.
Pricing source (opens in a new tab)

Data & enterprise terms

API Terms section 10 commits not to train on covered API inputs or outputs. The pricing page separately labels no-train as a Provisioned Throughput benefit and says no guarantee for pay-as-you-go; this conflict requires contractual confirmation.

Read the service policy (opens in a new tab)

Luma Agents API

Provider source (opens in a new tab)
API model ID
ray-3.2
Access
Documented API access; an eligible account, API key and billing setup are required. No paid generation was performed.
Regions
Not collected
API endpoint URL
POST https://agents.lumalabs.ai/v1/generationsSource for API endpoint URL (opens in a new tab)
API operation
video_reframeSource for API operation (opens in a new tab)
Asynchronous generation
SupportedSource for Asynchronous generation (opens in a new tab)
Download URL lifetime
1 hour; poll again to refreshSource for Download URL lifetime (opens in a new tab)
Video input duration
At most 10 secondsSource for Video input duration (opens in a new tab)
Input video limits
At most 200 MB for URL/base64 uploadsSource for Input video limits (opens in a new tab)
HDR output
Not supportedSource for HDR output (opens in a new tab)

SDR only; EXR also rejected.

Output resolutions
360p, 540p, 720p, 1080pSource for Output resolutions (opens in a new tab)

1080p portrait 9:16/3:4 is currently rejected as coming soon.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Video generation360p (draft) SDR1$0.03second
Video generation540p SDR1$0.06second
Video generation720p SDR1$0.12second
Video generation1080p SDR1$0.36second

Rate conditions

  1. video_reframe only. SDR. Bills per second; 1080p portrait targets are not yet supported. Published video rates are subject to change ahead of general availability.
Pricing source (opens in a new tab)

Data & enterprise terms

API Terms section 10 commits not to train on covered API inputs or outputs. The pricing page separately labels no-train as a Provisioned Throughput benefit and says no guarantee for pay-as-you-go; this conflict requires contractual confirmation.

Read the service policy (opens in a new tab)

Luma Agents API

Provider source (opens in a new tab)
API model ID
ray-3.2
Access
Documented API access; an eligible account, API key and billing setup are required. No paid generation was performed.
Regions
Not collected
API endpoint URL
POST https://agents.lumalabs.ai/v1/generationsSource for API endpoint URL (opens in a new tab)
API operation
video with exactly one generation_id start/end anchorSource for API operation (opens in a new tab)
Asynchronous generation
SupportedSource for Asynchronous generation (opens in a new tab)
Download URL lifetime
1 hour; poll again to refreshSource for Download URL lifetime (opens in a new tab)
HDR output
Not supportedSource for HDR output (opens in a new tab)

SDR only; billed as one 5-second block.

Prices retain their published currencies and billing units below.

Published billing units · original currency · observed 12 Oct 2026
CategoryUsageRateUnit
Video generation540p SDR1$0.15extension
Video generation720p SDR1$0.30extension
Video generation1080p SDR1$1.20extension

Rate conditions

  1. extend only. SDR. Always one 5-second extension block. Published video rates are subject to change ahead of general availability.
Pricing source (opens in a new tab)

Data & enterprise terms

API Terms section 10 commits not to train on covered API inputs or outputs. The pricing page separately labels no-train as a Provisioned Throughput benefit and says no guarantee for pay-as-you-go; this conflict requires contractual confirmation.

Read the service policy (opens in a new tab)

Caching

Caching can reuse earlier input processing. Availability, lifetime and billing depend on the endpoint. A cache hit still generates a new response.

01 / INITIAL INPUTProcess & retain

The system processes eligible input and retains reusable state.

02 / LATER REQUESTCheck available state

Reuse depends on matching input, session state and the cache policy.

03 / CACHE HITReuse & generate

Reusing state can reduce repeated work. Any charges follow the billing rules below.

Luma Agents API

ray-3.2
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Luma Agents API

ray-3.2
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Luma Agents API

ray-3.2
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

Luma Agents API

ray-3.2
Not verified
How it works
Not collected
Minimum reusable input
Not collected
Lifetime & refresh
Not collected
How to enable it
Not collected
Usage fields
Not collected
Billing rules
Not collected
Limits & data boundaries
Not collected

No verified cache rate for this endpoint. An empty rate is not a zero charge.

What the measurements say

No benchmark observations have been added for this model. We do not turn a vendor description into a performance score.

Latency and throughput depend on the host, request size, reasoning settings and load. There is no measured speed or global usage figure in this record.

The record behind the model

Why it is included

Inferred from the named Luma-hosted proprietary service, terms restricting access to underlying models and weights, and the separate official public-checkpoint inventory (IMM and TVM). This is release-specific catalog classification, not an assertion that all Luma research is closed.

Classification evidence (opens in a new tab)

Lifecycle dates

Released9 Jun 2026Source (opens in a new tab)

Sources reviewed