Luma AI

What Luma AI does

Luma AI is a frontier multimodal AI company founded in 2021 by former Apple AR and computer vision engineers, headquartered in Palo Alto with international offices in London, Riyadh, Munich, Singapore, and several US cities. The company builds its own foundation models for video (Ray3.2, Ray3.14, Ray3 Modify) and image (Uni-1, Uni-1.1) generation, anchored by proprietary training algorithms including Inductive Moment Matching and Terminal Velocity Matching, and a Unified Intelligence decoder-only autoregressive transformer architecture. Its core product, Luma Agents, orchestrates planning, generation, iteration, and refinement of creative work across video, image, audio, and text within a single workflow with shared context, distributed through the consumer Dream Machine app, enterprise sales motion, developer APIs, and channel partners.

The platform serves enterprise marketing and creative teams, advertising agencies (Publicis Groupe Middle East, Serviceplan Group, Dentsu), film and production studios (Wonder Project, Mazda), e-commerce brands, educators, and individual creators. Luma generates revenue through tiered SaaS subscriptions (Plus $30, Pro $90, Ultra $300 per month, plus Team and Enterprise plans), usage-based API pricing for both Ray3.2 video and Uni-1.1 image generation (with a dedicated Scale tier for SLA-backed capacity), and custom enterprise contracts including custom fine-tuning and Forward Deployed Creatives. The company has raised approximately $1.3B+ across seed, Series A, B, and a $900M Series C led by Saudi Arabia's HUMAIN at a ~$4B post-money valuation in November 2025, with the raise paired with a 2GW Project Halo supercluster joint venture with AMD, Cisco, and HUMAIN.

As of early 2026, Luma had grown to 150+ employees and 30M+ users worldwide, with annualized revenue of $8M reported as of June 2025. The company distributes through Adobe Firefly and Creative Cloud integrations, eight+ developer platform partners (Envato, Comfy, Fal, Runware, Krea, Flora, Magnific, LovArt), direct enterprise sales, the Open Physical AI Lab research initiative, and education programs at universities including USC, UC San Diego, Loyola Marymount, and Chapman. The company has made two disclosed acquisitions (Apparate Labs in December 2024 and Persona in Q1 2026) to extend platform capabilities.

Name

Luma AI

Legal name

Luma AI, Inc.

Website

https://lumalabs.ai

Company type

Private

Founded year

2021

Operating status

Operating

Headcount range

101–250 employees

Short description

Luma AI builds frontier multimodal foundation models and Luma Agents that generate, refine, and orchestrate video, image, audio, and text creative work for enterprise marketing teams, advertising agencies, film studios, developers, and individual creators.

Ownership category

akta.pro rank

Product category

Generative AI Video & Image Platform

NAICS

Software Publishers (513210), Custom Computer Programming Services (541511)

akta.pro primary industry

End-to-End Enterprise AI Platforms (MLOps & Model Lifecycle Management) (HDAEANAA)

akta.pro secondary industries

Model Serving, Inference & Deployment Platforms (APIs, Edge/On-Prem) (HDAEANAC), AI Integration & Orchestration Platforms (Connectors, Workflow, iPaaS for AI) (HDAEANAI), Enterprise AI Data & Knowledge Platforms (Vector Databases, Knowledge Graphs) (HDAEANAH)

Keywords

• AI video generation
• AI image generation
• Generative AI models
• Creative agent platform
• Multimodal foundation models

Headquarters

HQ city

Palo Alto

HQ country

United States

HQ region

North America

Offices9 records

Markets served

GTM type

B2B and B2C

Offering type

Software

Cost components

Personnel, Technology or R&D, Infrastructure, Marketing or Sales, Operations

Revenue model

1. Luma App Subscription Plans: Tiered SaaS subscriptions: Plus ($30/month), Pro ($90/month with 4x usage), and Ultra ($300/month with 15x usage), with annual billing offering ~17% discount. Includes access to Luma's own models and third-party image/video models, commercial use rights, and edit access for guest collaborators.
2. Ray3.2 API Pay-as-you-go: Usage-based API billing with no minimum commitment. T2V/I2V at 540p 5s = $0.15, scaling to 1080p 10s = $3.60. V2V, Reframe, and Modify priced similarly. HDR is 2x SDR, HDR+EXR is 3x SDR. Dedicated Scale tier available with SLAs.

Pricing tiers

Go-to-market motion1 record

Distribution channels11 records

Marketing channels12 records

Core offering

Luma AI develops and operates a creative agent platform that orchestrates video, image, audio, and text generation through its proprietary foundation models (Ray3.2 for video, Uni-1 for image) alongside integrated third-party models. The platform enables enterprise creative teams, advertising agencies, studios, developers, and individual creators to produce cinema-grade video, on-brand image assets, and end-to-end campaigns via web apps, a developer API, and enterprise deployments.

Product overview

Luma AI offers a unified multimodal AI platform anchored by Luma Agents, the core product that orchestrates specialized models across video, image, audio, and text within a single creative workflow with shared context. The platform is built on Luma's proprietary flagship models: Ray3.2 for video generation (the world's first reasoning video model with native HDR and EXR export, supporting up to 16 keyframes per clip and 1080p output up to 20 seconds) and Uni-1 for image generation (a decoder-only autoregressive transformer that unifies text and image reasoning). These models are complemented by the Dream Machine video platform, the Ray3.14 and Ray3 Modify variants for different video workflows, the Open Physical AI Lab research initiative for world models and robotics, and Luma Skills for repeatable creative workflows. The platform is accessible via consumer web apps (AI Video Generator, AI Video Generator from Image), the Dream Machine studio, and a developer API supporting Python, JavaScript/TypeScript, and Go SDKs with REST endpoints. Enterprise tiers (Luma Enterprise) add SSO, dedicated Forward Deployed Creatives, and custom fine-tuning, while specialized programs serve education (Luma for Education) and creators (Creative Partner Program). The platform also integrates third-party models including ElevenLabs for audio, Google Veo 3.1 for video, Kling for video, and GPT Image 2 for images.

Differentiator

Problem solved

Functional benefit

Brands

• Ray: Reasoning-driven video generation model family (Ray, Ray2, Ray3, Ray3.14, Ray3.2) used to generate cinematic video from text, image, and video inputs.
• Uni-1
• Luma Agents
• Dream Machine

Products and services

• Luma Agents Core platform product consisting of creative agents that plan, generate, iterate, and refine creative work end-to-end with shared context across teams, operating across video, image, audio, and text. Agents coordinate built-in editing and refinement within the same project, scale execution in parallel, and carry context from planning through final assembly.
• Ray3.2 Flagship video generation model with frame-level control via multi-keyframe (up to 16 keyframes per clip), 1080p output up to 20 seconds, native HDR Generation (10/12/16-bit), and 16-bit EXR Export for professional production pipelines. Supports text-to-video, image-to-video, video-to-video (Modify Video V2), and Reframe. Now available as a full API.
• Ray3 The world's first reasoning video model with an internal reasoning system that interprets prompts, infers intent, and plans how scenes should unfold before rendering. Achieves state-of-the-art performance alongside Veo3 in independent evaluations.
• Ray3.14 Next-generation video generation model offering native 1080p video output, 3x cheaper and 4x faster than prior Ray models, designed for professional animation and cinematic workflows with HDR Generation and EXR Export capabilities.
• Ray3 Modify Hybrid-AI model for enhancing and editing actor performances in videos within Dream Machine. Enables realistic manipulation of performances with control over motion, timing, and emotional delivery, addressing limitations of early AI video systems.
• Uni-1 Multimodal reasoning image model built on Unified Intelligence architecture. A decoder-only autoregressive transformer where text and image tokens share a single sequence as input and output. Ranks #1 on Human Preference Elo for Overall, Style & Editing, and Reference-Based Generation; #2 for Text-to-Image. Supports up to 9 reference inputs for character/style preservation.
• Uni-1.1 API REST API for the Uni-1.1 image generation model with Generate Image and Modify Image endpoints. Supports up to 9 reference inputs per request with natural-language editing. Already in production with partners including Envato, Comfy, Runware, Flora, Krea, Magnific, Fal, and LovArt.
• Dream Machine AI video generation platform that hosts the Ray model family including Ray3, Ray3.14, Ray3.2, Ray3 Modify, and Modify Video V2 capabilities. Combines photorealistic generation with professional-grade editing tools including precision frame refinement, smart erase & fill, subject-aware editing, and prompt editing.
• Luma API Developer API exposing the full control surface of Ray3.2 for video generation and Uni-1.1 for image generation. Available via REST with Python, JavaScript/TypeScript, and Go SDKs. Two tiers: Build (pay-as-you-go, no commitment) and Scale (dedicated capacity with SLA).
• Luma Enterprise Enterprise offering with dedicated Forward Deployed Creatives, enterprise-grade SSO, access controls, brand customization, unlimited generations across media types, and custom fine-tuning options for large organizations.
• AI Video Generator (Consumer Web App) Consumer-facing AI video generator web app that turns text descriptions, images, and prompts into cinematic production-quality video. Powered by Ray, providing native 1080p output with HDR support and EXR export. Free to try.
• AI Video Generator from Image Image-to-video generator powered by Ray, turning static images into cinematic video. Ray reasons about the scene (subject, depth, lighting) before rendering, producing 1080p output up to 20 seconds with physics-aware motion and cinematic camera control from plain language.

Quantifiable outcome

• AI tools reduced video production costs by 91%, from ~$4,500 to ~$400 per minute of professional video (industry-wide statistic cited by Luma)
• +6 more outcomes

Named customers22 records

Segments10 records

Ideal customer profiles6 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration8 records

AI capability14 records

Feature10 records

Partnerships

28 partnerships are on record, tiered tier-1, tier-2 and flagship / tier-1.

• Wonder Projecttier-1Strategic or Co-development Partner · 1 June 2026Launched Innovative Dreams, a production services company combining generative AI and traditional filmmaking, backed and supported by Amazon Web Services. Jon Erwin (Founder of The Wonder Project) publicly endorsed Luma as embedding generative AI directly into the production pipeline.
• Boundless (South African creative agency)tier-2Channel Partner/ Reseller/ Distributor · 1 April 2026Boundless produced Mazda's first AI-produced commercial using Luma Agents and signed an agreement to integrate Luma AI's technologies across future client work. Campaign for Mazda MX-5 re-entry into South African market completed in less than two weeks.
• Amazon Web Services (AWS)tier-1Technology or Integration · 1 April 2026AWS supports the Wonder Project and Luma's Innovative Dreams production services company, providing cloud infrastructure for combining generative AI and traditional filmmaking.
• Publicis Groupe Middle Eastflagship / tier-1Channel Partner/ Reseller/ Distributor · 1 February 2026Publicis Groupe Middle East selected Luma AI as its preferred generative AI technology partner for advertising and creative production across Saudi Arabia and the wider MENA region. The collaboration leverages Project Halo's 2GW AI supercluster to power next-generation generative AI applications at scale. Bassel Kakish (CEO Publicis Groupe Middle East & Turkey) publicly endorsed Luma.
• Serviceplan Grouptier-1Channel Partner/ Reseller/ Distributor · 1 February 2026Serviceplan Group has launched a strategic partnership with Luma AI to deploy AI for creative work across its global operations, making it the first of the world's largest agency groups to standardize AI for creative work at scale. Embeds AI into professional workflows spanning strategy, creative development, content production, and delivery across Serviceplan Group's 43 global locations and 6,500+ professionals.
• NVIDIAtier-1Strategic or Co-development Partner · 1 December 2025Nvidia-backed Luma AI announced $4B expansion in London. Project Halo partners with NVIDIA for AI infrastructure targeting up to two gigawatts of capacity by the early 2030s in collaboration with NVIDIA and AMD.
• WPP / Monks (Jason Day joined Luma from)tier-2Others · 1 December 2025Jason Day, formerly of WPP and Monks, was appointed as Head of EMEA at Luma AI to lead international business development from London, supporting growth plans across Europe, the Middle East, Africa, and Saudi Arabia.
• Cisco Systemstier-1Strategic or Co-development Partner · 1 November 2025Co-founded AI joint venture with AMD and Saudi Arabia's HUMAIN to build AI data centers in the Middle East. Launched with a 100-megawatt project in Saudi Arabia, with Luma as first customer. AMD and Cisco are minority shareholders sharing profits and losses while Humain leads. Plans to build up to one gigawatt of data centers by 2030.
• Adobeflagship / tier-1OEM/ Whitelabel/ Licensing Partner · 1 October 2025Ray2 integrated into Adobe Firefly and Firefly Boards (April 2025). Adobe MAX 2025 announced a dozen generative AI models available within Creative Cloud, including Luma AI's models alongside Google, OpenAI, and Runway. Luma AI models available in Firefly mobile app for iOS and Android. Adobe described Luma as a key integration partner.
• Oracle (Oracle Cloud Infrastructure / OCI)tier-2Technology or Integration · 1 October 2025Luma AI selected Oracle Cloud Infrastructure (OCI) for high-performance GPU clusters, low-latency networking, and reliable global infrastructure to support large-scale AI model training, inference, and deployment.
• Feature.iotier-2Strategic or Co-development Partner · 1 June 2025Strategic partnership with Luma AI to power next-gen personalized experiences in the engagement economy, announced June 2025.
• Freepik (now Magnific)tier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Envatotier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Faltier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Comfytier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Floratier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Runwaretier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Kreatier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Lovarttier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API. Production commitment for Uni-1.1 API.
• Weavytier-2Technology or IntegrationAPI partner building with Luma's Ray3.2 video generation API.
• ElevenLabstier-2Technology or IntegrationElevenLabs models (v3 text-to-speech, SFX v2 sound effects, Music v1 music generation) are integrated into Luma Agents as partner models for audio generation.
• Google (Veo 3.1)tier-2Technology or IntegrationVeo 3.1 video model integrated into Luma Agents as partner model for users to switch between models without leaving Luma.
• ByteDance (Seedance 2.0)tier-2Technology or IntegrationSeedance 2.0 video model integrated into Luma Agents as partner model.
• Kuaishou (Kling 3.0, Kling Omni)tier-2Technology or IntegrationKling 3.0 and Kling Omni video models integrated into Luma Agents as partner models.
• OpenAI (GPT Image 2)tier-2Technology or IntegrationGPT Image 2 and GPT Image 1.5 models integrated into Luma Agents as partner models.
• Google DeepMind (Nano Banana, Nano Banana Pro)tier-2Technology or IntegrationNano Banana and Nano Banana Pro image models integrated into Luma Agents as partner models.
• ByteDance (Seedream)tier-2Technology or IntegrationSeedream image model integrated into Luma Agents as partner model.
• Hartbeattier-2GTM or Marketing PartnerHartbeat and Luma AI partnered on Prompt Side Story, a first-of-its-kind live AI film event at LA Tech Week in October 2025.

Scale indicators14 records

Recent moves8 records

Expansion highlights7 records

Direct peers

• Runway: Runway is the closest direct competitor to Luma in AI video generation (Gen-3/Gen-4) with a similar creative-pro audience and enterprise motion. Both sell production-quality video models to agencies, studios, and brands, and both offer Hollywood-grade integration.
• OpenAI: OpenAI's Sora competes head-to-head with Ray3.2 in frontier text-to-video and is integrated alongside Luma inside Adobe Firefly and Luma Agents. Direct overlap in image generation (GPT Image 2) versus Uni-1 makes OpenAI a primary peer despite its broader scope.
• Google DeepMind: Google's Veo 3.1 competes directly with Ray3.2 in reasoning-driven text/image-to-video, and Nano Banana competes with Uni-1. Veo 3.1 is integrated as a switchable model inside Luma Agents, underscoring head-to-head product overlap.
• Pika: Pika is a venture-backed AI video generation startup targeting the same creator, marketing, and prosumer segments as Luma's Dream Machine. Comparable business model (subscription tiers, creative-led GTM) and product category make it a tight direct peer.
• Midjourney: Midjourney competes with Uni-1 in frontier image generation and has expanded into video (Midjourney Video), sharing the same creative-pro, agency, and studio target customers as Luma. Both compete on Human Preference Elo rankings.
• Stability AI: Stability AI is a foundation-model lab offering competing open/closed image, video, and multimodal models (Stable Diffusion, Stable Video) with an API and enterprise motion similar to Luma's. Direct overlap in generative video/image and developer ecosystem.
• Kuaishou (Kling): Kuaishou's Kling 3.0/Kling Omni compete with Ray3.2 in high-resolution text/image-to-video and are integrated inside Luma Agents as switchable models. Strong product parity and shared enterprise/creator GTM make it a primary international peer.
• ByteDance (Seedance / Seedream): ByteDance's Seedance 2.0 (video) and Seedream (image) compete with Ray3.2 and Uni-1 respectively, and are integrated inside Luma Agents. Comparable frontier model quality and shared target customers in marketing/entertainment.

Broad incumbents

• Adobe (Firefly): Adobe's Firefly and Creative Cloud suite embed Luma as a third-party model while also competing with proprietary Firefly Video/Image models. Adobe is both a distribution partner and a broad incumbent across the same creative professional customer base.

Others

• ElevenLabs: ElevenLabs is a strategic integration partner whose audio models (TTS, SFX, music) are embedded in Luma Agents. Not a direct video/image competitor, but a key adjacent frontier-AI peer that jointly defines Luma's multimodal creative stack.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat6 records

Key risks5 records

Key highlights7 records

Customer concentration

• Xx.com/lumalabsai
• GitHubgithub.com/lumalabs
• YouTubeyoutube.com/@LumaLabs
• LinkedInlinkedin.com/company/luma-ai
• Instagraminstagram.com/lumalabs
• TikToktiktok.com/@lumalabs

Compliance2 records

Revenue estimate

Valuation estimate

Number of profiles

Profiles2 records

Subsidiaries1 record

Funding overview

Funding rounds5 records

Investors15 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A1 record

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Luma AI