AD
Luma AI logo

Luma AI

FreemiumπŸ‡ΊπŸ‡ΈUnicornScaling Losses

Generate cinematic video and images from text or images

86

Overall score

48

Heat score

Pricing

Free$0/month
Lite$9.99/month
Standard$29.99/month
Plus$64.99/month
Pro$99.99/month
Premier$499.99/month
EnterpriseCustom

Technical Specs

Inputs

Text Prompt, Image Upload, Video Upload, Reference Frame, Camera Motion Instruction, Style Reference

Outputs

AI Video Clip, AI Image, 4K Upscaled Video, HDR Video, EXR File, Extended Video, Modified Video

AI Type

Multimodal

Model Architecture

Custom/Proprietary

Daily Prompts

N/A

Context Length

N/A

Output Quality

Accuracy

87%

Content

90%

Reasoning

85%

Company Profile

Company

Luma AI

Founded

2021

HQ

Palo Alto, California, USA

Employees

224

Total Raised / Total Funding

$1.07B

Revenue

$55M

Valuation

$4B

ARR

$55M

CEO

Amit Jain

Overview

Estimated Paid Users

500K

Current estimate

Total Earnings Till Date

$55M

+8.00% from last month

Market Share

17%

Current share

Average Session

18

Per active user

Hallucination Rate

14%

Model quality signal

Growth Rate

+2.49%

Monthly active users

Burn Rate

$48M

Total expenses / years active

Paid User Gain

+8.75%

Monthly paid user trend

Profit Analysis

-$242M

Total Loss

$312.4M

Total Profit

$0

Performance Metrics

Accuracy

87%

Context

90%

Reasoning

85%

Safety

82%

Benchmarks

No benchmark scores available.

Luma AI Models

Ray1

Type: Video

Description: Original generative video model powering Dream Machine at launch. Supported text-to-video and image-to-video at draft resolution with 5-second clip generation.

Architecture: Custom/Proprietary

Ray2

Type: Video

Description: Second-generation video model with 10x compute over Ray1. Supports 1080p native output, 10-second clips, natural physics simulation, and coherent motion. Available on Amazon Bedrock and Dream Machine.

Architecture: Custom/Proprietary

Ray-Flash 2

Type: Video

Description: Lighter, faster variant of Ray2 designed for high-volume API usage. Optimized for rapid iteration and cost-efficient generation at scale for developers.

Architecture: Custom/Proprietary

Photon

Type: Image

Description: Luma's dedicated image generation model. Learns lighting, perspective, and photorealism from training data to deliver consistent still images that serve as the foundation for video creation in Dream Machine.

Architecture: Custom/Proprietary

Ray3

Type: Video

Description: World's first reasoning video model. Evaluates and refines its own outputs in real time. Supports native 1080p, 4K upscaling, true 16-bit HDR EXR video, Draft Mode, Hi-Fi Diffusion, and improved instruction following. Available in Dream Machine and Adobe Firefly.

Architecture: Custom/Proprietary

Funding Rounds & Investors

Total Funding

$1.07B

Rounds

0

No funding rounds available.

Founders/Team

AJ

Amit Jain

Co-Founder & CEO

AT

Alberto Taiuti

Co-Founder

AY

Alex Yu

Co-Founder & Former CTO

Direct competitors

No direct competitors available.

Change Log / Major Updates

2025 Β· Jan 15

Ray2 launches with 10x the compute power of Ray1, delivering 1080p native output, 10-second clips, improved motion coherence, and natural physics. Ray2 is made available on Amazon Bedrock and via the Dream Machine platform.

2025 Β· Sep 18

Ray3 is unveiled as the world's first reasoning video model, capable of evaluating its own outputs and refining them in real time. Ray3 introduces true HDR video in 16-bit EXR ACES2065-1 format. Adobe Firefly becomes first external partner to launch Ray3 outside Dream Machine. Dream Machine reaches 30M+ users.

2025 Β· Dec 19

Ray3 Modify launches, enabling hybrid-AI workflows where human actor performances can be transformed using AI while preserving motion and emotional delivery. Series C of $900M led by HUMAIN values Luma at $4B+. London office opens.

Compliance, Integrations & Support

Industry: Media Entertainment

Compliances: Not specified

Integrations: Adobe Firefly, Amazon Bedrock, AWS, NVIDIA, AMD, Snapchat, Unity, Dream Machine API, Photon Image API, HUMAIN Create, iOS App

Support:email, help center, community forum, enterprise support

Target audience: Content Creators, Filmmakers, Marketing Agencies, Indie Game Developers, Advertising Professionals, Social Media Creators, VFX Artists, Brands, Enterprises, Developers

Supported languages: English, Spanish, French, German, Japanese, Chinese, Portuguese, Italian, Korean, Arabic

Luma AI Acquisitions

No acquisition records available.

AD

Reviews & Rating

0 reviews

No reviews yet

Be the first to share how Luma AI performs for your workflow.

0.0

Accuracy

0.0

Ease of Use

0.0

Output Quality

0.0

Security

0.0

More About Luma AI

The Startup That Taught AI to Understand Physics

In September 2021, three engineers who had worked on Apple Vision Pro and AR systems walked out of Cupertino with a question nobody else was asking: what if AI didn't just generate pixels, but actually understood the world those pixels depicted? Amit Jain, Alex Yu, and Alberto Taiuti founded Luma AI to answer it, beginning with a deceptively humble tool β€” photorealistic 3D capture from iPhone footage.

Three years later, that philosophical bet has exploded into one of the most dramatic growth stories in generative AI. When Luma launched Dream Machine in June 2024, the platform signed up one million users in its first four days. By late 2025, over 30 million creators were using it globally to produce cinematic video from nothing but words and still images. The company's flagship models β€” Ray2 and Ray3 β€” are now embedded into Adobe Firefly, used by Hollywood VFX studios, and deployed via API at thousands of developer applications worldwide.

World Models, Not Just Pixels

What separates Luma from competitors is an architectural bet on world models: multimodal AI systems trained on video, audio, and language simultaneously. Where many video generators hallucinate physics β€” objects passing through walls, shadows moving in the wrong direction β€” Ray3 was engineered to reason about geometry, lighting, and motion before generating a single frame. Ray3, launched in September 2025, became the world's first reasoning video model capable of evaluating its own outputs and iterating in real time. It also introduced studio-grade HDR video in native 16-bit EXR format, a first for any AI system.

  • Ray3.14 β€” 4x faster, 3x cheaper than Ray3, with native 1080p generation
  • Ray3 Modify β€” hybrid AI workflow preserving human actor performance while transforming scenes
  • Photon β€” Luma's dedicated image generation model powering Dream Machine's still-image pipeline

In November 2025, Luma raised $900 million in Series C funding led by HUMAIN (Saudi Arabia's Public Investment Fund), bringing total funding past $1 billion and valuing the company at over $4 billion. The round will fund Project Halo β€” a 2-gigawatt AI supercluster in Saudi Arabia β€” and accelerate Luma's expansion into simulation, robotics, and personalized education at a global scale.

Luma AI FAQ's

What is Luma AI's Dream Machine?

Dream Machine is Luma AI's flagship platform for generating AI-powered videos and images from text prompts, image uploads, or existing video. It runs on Luma's proprietary Ray models and is available via web browser and iOS app.

What is the difference between Ray2 and Ray3?

Ray2, launched in January 2025, delivered 10x compute improvement over Ray1 with 1080p output and natural motion. Ray3, launched September 2025, adds reasoning capabilities β€” it can evaluate and refine its own outputs β€” plus true HDR video in 16-bit EXR format and improved instruction following.

Can I use Luma AI videos commercially?

Commercial use is permitted on the Standard plan ($29.99/month) and above. The Free and Lite plans produce watermarked content restricted to personal use only.

How does the credit system work?

Each plan comes with a monthly allotment of credits that refresh each billing cycle. Video generation costs vary by resolution and duration β€” for example, a 10-second Ray3 clip at 720p SDR costs 640 credits. Credits do not roll over, but you can purchase Top-Up Credits any time on paid plans.

Is Luma AI's API separate from Dream Machine subscriptions?

Yes. Dream Machine subscription credits cannot be used for API calls. API access requires separate credit purchases via the Dream Machine API billing portal, and is priced independently.