Luma AI pricing

About

Luma AI builds Dream Machine, a generative-media platform that turns text and images into video, generates and edits images, and produces audio — surfacing Luma’s own models (Ray3.2, Ray3.14, the Uni-1 image model) alongside a roster of third-party models (Veo 3.1, Veo 3, Kling, Seedance, GPT Image, Nano Banana, Seedream, ElevenLabs) inside a single workspace. Beyond video, Luma is known for Genie, its text-to-3D generation product, and for NeRF/Gaussian-splatting research that put the company on the map before Dream Machine. The product targets individual creators, studios and agencies, with a Team plan for shared workspaces, an Enterprise tier, and a developer API for embedding the models.

Luma monetizes through per-seat Individual subscriptions (Plus, Pro, Ultra) that each bundle a monthly credit allowance, with credits acting as the metered unit across every generation type. Two Business plans (Team, Enterprise) are sales-led and quoted, adding shared team credits, SSO, enterprise commitments and custom fine-tuning. A separate pay-as-you-go API prices generations directly in dollars. Company financials (ARR, headcount, valuation) are left for wiki-research.

For the most current information on Luma AI’s pricing and market position, visit Luma AI.

Pricing summary : How Luma AI’s pricing model works

Luma runs a seat-plus-credits model on Dream Machine. You buy a per-seat Individual subscription — Plus $30/month, Pro $90/month, or Ultra $300/month (each advertising “Save up to 20% with yearly” billing: $25/$75/$250 per month billed $300/$900/$3,000 yearly) — and each plan bundles a monthly pool of credits that is the single metered unit consumed by every video, image and audio generation. Luma now posts an explicit monthly credit total on each plan card — Plus 10,000 credits, Pro 40,000 credits, Ultra 150,000 credits — while still framing the step-up with a relative usage multiplier (Pro “4x usage with the Luma Agents”, Ultra “15x usage”). Team and Enterprise are “Contact us” Business plans adding shared team credits, usage analytics, SSO, enterprise commitments and custom fine-tuning.

Billing dimensions captured:

• Per-seat Individual subscription — Plus $30/month, Pro $90/month, Ultra $300/month; each advertising “Save up to 20% with yearly” billing ($25/$75/$250 per month, billed $300/$900/$3,000 yearly).
• Posted monthly credit allowance — Plus 10,000 credits, Pro 40,000 credits, Ultra 150,000 credits, printed directly on each plan card.
• Relative usage multipliers — Pro is “4x usage with the Luma Agents”, Ultra is “15x usage”, shown alongside the credit total.
• Credits — a single metered unit consumed per generation, priced per model and resolution (see the per-model grid below); subscription credits do not roll over.
• Commercial use — an explicit feature starting on Plus and carrying through Pro and Ultra.
• Business plans — Team and Enterprise quoted, adding shared team credits, usage analytics, SSO, enterprise commitments and custom fine-tuning.
• Pay-as-you-go API — generations billed directly in dollars per generation (see Hidden costs for representative rates), with HDR output at 2x and HDR+EXR at 3x.

What makes this different: Luma sells seats but meters everything in one credit currency whose per-generation cost is published as a dense, model-by-model grid — a textbook credit-based billing layer wrapped around a seat-plus-usage subscription. Each plan now shows both an absolute credit number (10,000 / 40,000 / 150,000) and a “4x / 15x usage” multiplier, so buyers can size an allowance against the per-generation grid instead of reverse-engineering it from throughput framing alone.

Pricing by product

Dream Machine (subscription plans)

Dream Machine API (usage-based)

The developer API prices generations directly in dollars per generation rather than plan credits, scaling with model, resolution and duration. HDR output is 2x SDR pricing and HDR+EXR is 3x. Representative computed dollar rates are shown in the Hidden costs section below.

Sales motions across products: PLG / self-serve for the Plus, Pro and Ultra Individual plans and the pay-as-you-go API; sales-led for the Team and Enterprise Business plans (shared credits, SSO, enterprise commitments and custom fine-tuning are quote-stage). The free/draft entry point sits at the top of the self-serve funnel into paid seats.

Hidden costs : What Luma AI users actually pay

The sticker price on a Dream Machine plan is the seat fee, but the number that actually moves your spend is how fast your credit pool drains — and the per-generation cost swings enormously by model and resolution. Each plan now posts its monthly credit total (Plus 10,000, Pro 40,000, Ultra 150,000), so the allowance is known up front; the only way to tell whether that allowance covers your work is still to map your generation mix onto the credit grid. The figures below are computed from Luma’s published per-generation credit costs and API dollar rates, so treat the dollar-per-output numbers as approximate.

Archetype A — credit cost per video clip varies ~100x by model. A creator on a fixed monthly pool pays very different credit prices for the same 5-second 720p clip depending on the model picked.

Archetype B — images and audio quietly draw from the same wallet. The “video” tool also meters images and audio, and a single video clip can cost as much as a hundred images.

Other things to budget for: subscription credits do not roll over — unused allowance resets each billing cycle, so bursty project work wastes part of the pool. HDR and EXR output multiply cost (2x and 3x SDR) on the API and consume proportionally more credits in-app. The per-plan credit total is now posted (Plus 10,000, Pro 40,000, Ultra 150,000), so you can size a tier without reverse-engineering the “4x / 15x usage” multiplier — but the total still only self-qualifies once mapped to your actual model mix. Commercial use is gated to paid plans, so any professional output effectively requires at least Plus.

Want to estimate your own Luma AI bill? Use the Luma AI pricing calculator to model your monthly cost based on plan, generation volume and model mix.

Pricing evolution : Luma AI pricing history and changes

Cadence

Tracked range: 2024–present. Between the 2024 launch (credit-total framing) and the 2026-06-11 capture (multiplier-only framing), intermediate plan changes are not individually snapshot-verified on this network; the 2026-07-23 capture confirms credit totals are posted again alongside the multiplier.

Notable changes

• 2024-06 — Dream Machine launches as a freemium product: a small monthly free credit allowance at draft resolution plus paid plans, establishing the credit-metered generation model. Early tiers (Lite/Plus/Unlimited) were framed directly by monthly credit totals.
• By 2026-06 — Individual ladder is Plus $30 / Pro $90 / Ultra $300 (up to 20% off annually), with tiers at this point framed by a relative usage multiplier (“4x usage with the Luma Agents” on Pro, “15x” on Ultra) and no posted credit number (source: captured lumalabs.ai/pricing, 2026-06-11).
• By 2026-06 — Two Business plans (Team, Enterprise) are quoted, adding shared team credits, usage analytics, SSO, enterprise commitments and custom fine-tuning; a separate pay-as-you-go API prices generations in dollars with HDR (2x) and HDR+EXR (3x) multipliers.
• 2026-07-23 — Luma re-posts explicit per-plan credit totals — Plus 10,000, Pro 40,000, Ultra 150,000 credits — printed on each Individual plan card alongside the existing 4x/15x multiplier, ending the multiplier-only framing; headline prices are unchanged ($30/$90/$300; $25/$75/$250 billed yearly). The per-model credit grid also expanded with newer models (Seedance 2.0/Mini, Kling Omni/3.0/2.6, Veo 3.1/3, Nano Banana 2/Pro, GPT Image 2/1.5) (source: captured lumalabs.ai/pricing, 2026-07-23).

What’s unique : Luma AI’s distinctive pricing mechanics

1. Tiers carry two sizes at once — an absolute credit total and a relative multiplier.

As of the 2026-07-23 capture, each Individual plan card prints both an explicit monthly credit total (Plus 10,000, Pro 40,000, Ultra 150,000 credits) and a relative usage multiplier (Pro “4x usage with the Luma Agents”, Ultra “15x usage”). The two are the same number said twice — 40,000 is exactly 4x of 10,000 and 150,000 is exactly 15x — so the multiplier is the credit total restated against the entry tier. Posting the absolute allowance (which the earlier 2026-06 cards withheld in favor of the multiplier alone) lets buyers size a plan directly against the per-generation grid, while the multiplier keeps a quick “how much more headroom” read for shoppers who don’t want to do credit math.

2. One credit currency spanning video, image and audio.

Every modality — Ray/Veo/Kling video, GPT Image/Nano Banana/Seedream images, ElevenLabs speech, sound effects and music — is metered in the same credit unit. A 1-credit Seedream image and a 140-credit Veo 3.1 clip both draw from one wallet, so the metered unit means radically different things depending on what you generate. This is credit-based billing stretched across three modalities at once.

3. A marketplace of third-party models priced in Luma’s credits.

Dream Machine surfaces Veo, Kling, Seedance, GPT Image, Nano Banana and ElevenLabs alongside Luma’s own Ray and Uni-1 models — all billed in Luma credits at model-specific rates. Luma effectively runs a re-priced model marketplace where the credit becomes a universal accounting layer over a portfolio of external inference costs it can mark up or subsidize per model.

4. Resolution and HDR as explicit price multipliers.

Cost scales with output fidelity: higher resolution costs more credits, and HDR output is 2x SDR pricing while HDR+EXR is 3x on the API. Fidelity is a first-class pricing axis, so the same clip can cost 4x more at 1080p HDR than at 540p SDR — a usage-based lever that ties price directly to compute.

Strengths & weaknesses

Billing UX : Luma AI billing controls and transparency

• Monthly / Yearly toggle — the pricing page exposes a Monthly vs Yearly switch advertising “Save up to 20% with yearly”, showing both per-plan rates before purchase.
• Per-model credit grid — Dream Machine publishes a detailed cost-per-generation grid (video by model/resolution/audio, image by model/quality, audio by type) so users can see credit cost before generating.
• Posted credit totals + usage multiplier — paid plan cards now show both an explicit monthly credit allowance (Plus 10,000, Pro 40,000, Ultra 150,000) and a relative usage multiplier (“4x” / “15x”), so buyers can read absolute allowance and relative throughput side by side.
• Shared team credits & analytics — the Team plan pools credits across members and adds usage analytics plus SSO, moving billing controls to the workspace level.
• API dollar pricing — the developer API surfaces explicit per-generation dollar rates (with HDR/EXR multipliers), so usage-based spend is priced in dollars rather than credits.
• Enterprise quote flow — Team and Enterprise route to a “Contact us / Get in touch” form for shared credits, enterprise commitments and custom fine-tuning.

Strategic wins : Why Luma AI’s pricing decisions worked

1. One credit ledger over a multi-model marketplace.

By metering Luma’s own models and a roster of third-party models (Veo, Kling, ElevenLabs) in a single credit currency, Luma can add, remove or re-price any model without changing its plans. The credit becomes a universal accounting layer over a portfolio of external inference costs — exactly the kind of decoupling that lets compute-heavy AI products shift away from per-user licenses toward consumption metering.

2. Bundling commercial rights from the entry tier.

Putting commercial use on the $30 Plus plan (not gating it to an expensive top tier) removes a major adoption barrier for working creators and agencies. It converts free/draft evaluators into paying customers the moment they need to ship — a clean monetization trigger. See choosing the right usage metric.

3. A self-serve API alongside the seat app.

The pay-as-you-go API, priced per generation in dollars with HDR multipliers, lets developers embed Luma’s models without a sales call — running the same models through two go-to-market motions (PLG seats for creators, usage-based dollars for developers) off one inference platform. Related: outcome-based pricing trends.

Areas to improve : Gaps in Luma AI’s pricing approach

1. Translate the posted credit total into expected output.

Luma closed the biggest gap here on 2026-07-23 by printing an explicit monthly credit number on each plan card (Plus 10,000, Pro 40,000, Ultra 150,000) — a real self-qualification win over the earlier multiplier-only framing. The next step is helping buyers convert that number into work: a per-plan “your 10,000 credits is roughly X seconds of Ray3.14 video or Y Seedream images” estimator would turn an abstract allowance into a decision, since a raw credit count still means little until it is mapped onto a per-generation grid where cost swings ~100x. See bill shock and cost unpredictability.

2. Add credit roll-over or expiry warnings.

Because subscription credits reset each cycle and do not roll over, the lumpy, project-based workflows creators actually have waste part of the pool every quiet month. A modest roll-over cap or an in-product “credits expiring soon” nudge would soften the all-or-nothing reset. Tools like thresholding and alerting on usage are the standard pattern.

3. Tame the ~100x per-model cost spread.

A single credit buying anywhere from one image to a fraction of a Veo clip makes the wallet hard to reason about. Clearer in-product “this generation will cost N credits / you have M left” framing — and a simple cost-per-second-of-output comparison across models — would reduce the unpredictable-cost anxiety that pushes creators to over-buy.

Key takeaways

1. Meter in one credit currency across modalities. Luma runs video, image and audio through a single credit pool over many models — a flexible accounting layer that lets it re-price inference without touching plans.
2. Post the absolute allowance and the relative multiplier together. Luma initially sold Pro as “4x” and Ultra as “15x usage” with no credit number, then re-added explicit totals (10,000 / 40,000 / 150,000) on 2026-07-23; showing both lets shoppers read relative headroom at a glance while giving serious buyers the absolute number they need to self-qualify.
3. Bundle commercial rights early to convert. Commercial use on the $30 entry plan turns free evaluators into payers the moment they need to ship — a clean monetization trigger.
4. Per-generation cost can swing 100x. The same credit buys a 1-credit image or a slice of a 140-credit Veo clip, so credit budgets only mean something once mapped to a specific model mix.
5. Run one inference platform through two motions. Seat-based app for creators plus a pay-as-you-go API for developers lets Luma serve both buyers off one metering system.

UBP implications

1. Credit currencies let GPU-bound businesses re-price the cost-of-goods layer continuously. Because Luma meters in abstract credits and only the per-model conversion changes, it can absorb shifts in inference cost (new models, third-party rate changes) without re-papering plans — a structural advantage of credit-based billing. See usage-based pricing strategy.
2. A usage multiplier is a packaging layer, not a substitute for the allowance. Luma’s 2026-07-23 move — re-adding absolute credit totals (10,000 / 40,000 / 150,000) beside its “4x / 15x usage” multiplier — shows the trade-off resolving in the buyer’s favor: the multiplier is marketing-friendly, but only the posted number lets buyers self-qualify, and carrying both costs nothing while closing the opacity gap.
3. Fidelity (resolution, HDR) is an underused price axis. Tying credit cost to output quality (HDR 2x, EXR 3x) ties price directly to compute and lets a single product span casual and professional spend without separate plans.

Sources

• Luma AI pricing page (accessed 2026-06-11) — live capture: Plus $30, Pro $90, Ultra $300; Team & Enterprise “Contact us”; per-model credit grid
• Luma API pricing (accessed 2026-06-11) — per-generation dollar rates; HDR 2x, HDR+EXR 3x
• Dream Machine plans & credits — Luma Learning Hub (accessed 2026-06-11) — credits do not roll over; historical Lite/Plus/Unlimited credit totals
• Luma Dream Machine pricing 2026 — magichour.ai (accessed 2026-06-11) — third-party plan & free-tier summary
• Luma Dream Machine pricing breakdown — PhotonPay (accessed 2026-06-11) — third-party tier & hidden-fee summary

Bottom line

Luma AI prices Dream Machine as a clean per-seat ladder — Plus $30, Pro $90, Ultra $300 per month (up to 20% off annually) — wrapped around a single credit currency that meters video, image and audio across both Luma’s own models and a marketplace of third-party ones. As of the 2026-07-23 capture each plan card posts both an explicit monthly credit total (10,000 / 40,000 / 150,000) and the older “4x / 15x usage” multiplier, so buyers can now read prices and allowance clearly; the residual friction is downstream — a credit only means something once mapped onto a per-generation grid where cost swings nearly 100x by model. Commercial rights bundled from the entry plan and a self-serve pay-as-you-go API make Luma a strong case study in how far a multi-modal credit model stretches once the allowance is on the table and the remaining work is output-level predictability. Browse the pricing blueprint for more fully-researched company profiles.

Want to compare Luma AI against other generative-media and AI-platform companies? Browse the pricing blueprint.

Pricing timeline : Major events on a vertical axis

Each milestone below corresponds to a public pricing change, product launch, or material adjustment. Major events use a filled marker; minor adjustments use a faded one.

Per-plan credit totals now posted (Plus 10,000 / Pro 40,000 / Ultra 150,000)

Jul 2026

Captured lumalabs.ai/pricing (USD): prices unchanged (Plus $30, Pro $90, Ultra $300/mo; $25/$75/$250 billed yearly), but each Individual plan card now prints an explicit monthly credit allowance — Plus 10,000, Pro 40,000, Ultra 150,000 credits — alongside the existing 4x/15x usage multiplier, ending the earlier 'no posted credit total' framing. The per-model credit grid also expanded with new models (Seedance 2.0/Mini, Kling Omni/3.0/2.6, Veo 3.1/3, Nano Banana 2/Pro, GPT Image 2/1.5).

Plus $30 / Pro $90 / Ultra $300 Individual plans + multiplier usage + API

Jun 2026

Captured lumalabs.ai/pricing (USD): Individual plans Plus $30/mo, Pro $90/mo (4x usage with the Luma Agents), Ultra $300/mo (15x usage), each up to 20% cheaper yearly. Business plans Team and Enterprise are 'Contact us' (shared team credits, SSO, custom fine-tuning). Tiers were framed by relative usage multipliers rather than posted credit totals; a detailed per-model credit cost grid (video/image/audio) and a separate pay-as-you-go API sit alongside.

Dream Machine launches with a free credit allowance

Jun 2024

Luma launched Dream Machine (text/image-to-video) with a freemium model — a small monthly free credit allowance plus paid plans — establishing the credit-metered generation model that still underpins pricing today. Early plan tiers (Lite/Plus/Unlimited) were framed directly by monthly credit totals.