> ## Content Index > Fetch the complete content index at: https://www.theaidirectory.com.au/llms.txt > Use this file to discover other available public pages before exploring further. # Lovart Review (2026): The AI Design Agent, Tested for Australia - URL: https://www.theaidirectory.com.au/review-lovart-australia-2026/ - Published: 2026-06-24T23:00:00.000Z - Updated: 2026-10-09T06:32:43.000Z - Description: Lovart calls itself the world's first design agent — describe a brand and it produces logos, posters and a full kit on one canvas. Genuinely impressive output, brutal credit burn, and a US data story Australians need to read before they commit. - Author: Adam - Tags: Design & Creative, Lovart, #review **15**/25 ★★★☆☆ Good Lovart AI Verified 30 Sep 2026 Reviewed 25 Jun 2026 Value for Money**2**/5 Ease of Use**4**/5 Feature Depth**4**/5 Australian Fit**2**/5 Reliability & Support**3**/5 From **A$27**/mo · ex-GST · billed in USD (US$19) **The gate:** Five stars need 21+ and Australian Fit of at least 4/5. [Skip to the verdict ](#verdict) *Pricing figures updated 6 September 2026: USD prices are converted at the current rate (1 USD ≈ 1.40 AUD, verified 5 September 2026) and shown ex-GST, replacing the retired 1.55 + GST basis. A GST-registered Australian business that supplies its ABN is not charged GST on an imported digital service; other buyers should allow \~10% on top (*[*Methodology*](https://www.theaidirectory.com.au/methodology/)*).* > **The verdict:** Lovart is the most ambitious thing in AI design right now, and on a good run it is genuinely startling: you describe a brand in a sentence, and it plans the job, splits it into a task chain, and lays a full set of logos, posters, social tiles and storyboards across an infinite canvas — front to back, in one go. It works because it is not one model but an orchestra of them, routing each task to whichever engine suits it. The catch is the catch of the whole agentic-design category, only sharper here: the credit meter runs hard, a single complex brand job can eat hundreds of credits, unused credits expire monthly, and when an autonomous agent goes off-brief you spend more credits steering it back. Pricing is US dollars only, there is no Australian data region, and the support record has real blemishes. For ideation, mood-setting and first drafts it is exhilarating. For locked-down brand work on a budget, go in with your eyes open. Score: 15/25 — Good. ## What it is Lovart bills itself as **"the world's first AI design agent."** The distinction it is drawing is real. Most AI design tools are *generators* — you prompt, you get an image, you prompt again. Lovart is an *agent*: you give it a goal ("a brand identity for a Melbourne coffee roaster"), and it **decomposes the task itself**, plans the steps, and produces a whole coordinated design set — logo variations, a colour palette, posters, social media graphics, packaging mockups, even storyboards — arranged on an **infinite canvas** you can then edit with Photoshop-style layer controls. It is built by **LiblibAI**, a Chinese AI-image startup founded in May 2023 that runs one of the largest model-creation communities in the world. Lovart launched in **closed beta on 13 May 2025** — more than 20,000 designers joined the waitlist, and invite codes were briefly reselling on the grey market — before moving from beta to a general, commercial release. The service itself is operated by **Resonate International INC**, a Delaware-registered entity, which matters when we get to where your data lives. The headline pitch, then, is "design team in a box." The reality, as always, is more interesting and more conditional than the pitch. ## What it costs in Australia Lovart bills in **US dollars** on a **credit** model. There is a free plan, then four paid tiers (USD, monthly rate shown, re-read 30 September 2026; billed annually they are US$16, US$27, US$45 and US$109 a month, the last two on a promotion that renews at US$74 and US$166): | Plan | USD/mo | Approx AUD/mo (ex-GST) | Monthly credits | | -------- | ------ | ---------------------- | ---------------------------------------- | | Free | $0 | $0 | Daily allowance (\~30/day, no roll-over) | | Starter | US$19 | \~A$27 | 2,000 | | Basic | US$32 | \~A$45 | 3,500 | | Pro | US$90 | \~A$126 | 11,000 | | Ultimate | US$199 | \~A$279 | 27,000 | The credit model is the whole game, so read it closely. **Credits are not images** — they are a unit of compute, and how many a single action burns depends on what you ask for: the model selected, the image size and quality, the style and detail, and any advanced settings. A simple draft is cheap; a high-resolution, high-detail brand set is not. Because Lovart is an *agent* running many steps autonomously, one instruction can quietly trigger a chain of generations — and a chain of credit spend. **Credits reset monthly and do not roll over** — unused credits expire at the end of your billing cycle. Top-up credit packs you buy separately *do* persist and never expire, which is the one piece of good news in the model. Higher tiers (Basic, Pro, Ultimate) add an **"Unlimited Relax Generation"** mode — slower, queued generations on selected models that do not consume your fast credits — which softens the burn if you are not in a hurry. A few Australian specifics. Pricing is **USD only**; Lovart publishes no AUD price and no GST line. The AUD figures above are our own conversions at roughly 1.40 to the dollar, shown ex-GST — not quoted prices, so treat them as a guide and confirm the final charge at checkout. The weak Australian dollar adds about 40% over the sticker at 1.4045 (Frankfurter/ECB, 18 September 2026) before any tax. There is no Australian billing entity; charges flow through Lovart's US operator and its third-party payment processors. ## How it works / features You describe the outcome; Lovart plans and produces it. What you get: - **Agentic task planning.** Lovart breaks your brief into a visualised task chain and executes it across the canvas, producing multiple coordinated outputs from a single instruction rather than one image at a time. - **Infinite canvas with layer editing.** Outputs land on a zoomable canvas with Photoshop-grade controls — adjust text size, colour, and element positioning directly, rather than re-prompting from scratch. - **Multi-model routing.** This is the real engine. Lovart routes across a stable of leading models — **Flux** and **GPT Image** for stills, **Seedream**, **Nano Banana** and **Stable Diffusion** for different styles, and video models including **Veo 3, Kling, Sora and Hailuo** — and picks (or lets you pick) the right one per task. Sora is no longer part of that stable: OpenAI shut its API down on 24 September 2026\. You are effectively buying access to many paid models through one subscription. - **Full design sets.** Logos and brand kits, posters and ads, social media graphics, packaging mockups, and storyboards — the output is breadth, not a single asset. - **Export formats.** JPG, PNG and **SVG** (vector), the last of which matters for logos and print. - **Brand kits.** Paid tiers let you save brand kits (colours, fonts, logos) to keep generated sets consistent. - **Device limits.** Accounts are single-user — up to two desktop web sessions and one mobile web session at once; multi-user needs the separate Team Plan. On **ownership**: Lovart's terms state that you retain ownership of the content you create, and paid plans include **commercial usage rights** with no attribution required — usable for marketing, client work and merchandise. As with every generative tool, you also grant Lovart a licence to process your content to run and improve the service, and you are responsible for making sure your prompts do not infringe someone else's rights. For anything you intend to trademark or build a brand on, that last point is the one to take seriously — AI-generated logos sit in unsettled copyright territory worldwide. ## What it does well **The agentic workflow is the real thing.** When it lands, watching Lovart turn one sentence into a coherent, on-theme design set is the closest the category has come to "describe it and it appears." For ideation, mood-boarding and getting a first round of options in front of a client fast, it is genuinely ahead of single-image generators. **Many models, one bill.** Routing across Flux, GPT Image, Seedream, Stable Diffusion and the major video models means you are not paying four separate subscriptions to compare engines. For a creative who wants range, that consolidation has real value. **SVG export and layer editing.** Vector output and on-canvas editing put it ahead of tools that only hand you a flat PNG. You can take a logo to print, and you can fix a colour without re-rolling the whole generation. **A generous free plan to test on.** The daily free-credit allowance is enough to genuinely try the agent before you commit a cent — which, given the credit economics below, is exactly what you should do. ## Where it falls short **Credit burn is the defining problem.** Because Lovart runs autonomously across many steps, a single complex brand job can consume hundreds of credits — users report a full brand project eating 500-plus credits at a sitting, which makes even the 11,000-credit Pro plan feel thin by month's end. Credits expire monthly. Heavier users report real monthly spend drifting to US$100–150 once top-up packs are factored in. Budgeting actual cost is hard, and the meter rewards restraint you do not always have when an agent is doing the steering. **Autonomy cuts both ways.** The same agentic behaviour that produces a full set in one shot also means the agent sometimes goes off-brief — wrong tone, wrong layout, an unwanted element — and correcting it costs more generations, and more credits. For tightly specified, locked-down brand work where every pixel matters, a fully autonomous agent can be less efficient than a tool that does exactly and only what you tell it. Reviewers consistently note that precise, repeatable, on-spec output is where it is weakest. **No Australian data residency.** Lovart's privacy policy states data is stored on **servers in the United States**, with cross-border transfers to affiliates and service providers elsewhere. There is **no Australian region**, and the operator is a US (Delaware) entity with a Chinese parent in LiblibAI. The policy is framed around GDPR and US (CCPA/CPRA) law, with no specific Australian Privacy Act commitment. For anything touching Australian personal or client-confidential data, that is a material gap to weigh. **A patchy support and billing record.** Public reviews include real complaints — users locked out of paid accounts, cards charged without access, Pro credits not delivered as promised, and slow or absent support responses. We flag this not as a verdict on every account, but because the pattern appears often enough that an Australian buyer paying in USD, with no local entity to escalate to, deserves to know before subscribing. **Maturity.** Lovart only moved out of beta in the last year. It is improving fast, but it is a young product from a young company, and the polish and reliability of a mature design suite are not yet there. ## Australian fit **Pricing:** US dollars, no AUD shown and no GST line of Lovart's own — expect the exchange-rate premium plus likely GST through its payment processor, confirmed at checkout. **Data residency:** US servers, no Australian region, US operator with a Chinese parent, and no Australia-specific privacy commitment — a real consideration for Privacy Act-sensitive or client-confidential work. **Support:** no Australian entity, presence or local-time support, and a public record of billing and access complaints with no local body to escalate to. **Availability:** globally accessible, Australians can sign up and pay freely, and the free tier makes it easy to trial — but there is nothing Australia-specific in the product, the pricing or the data handling. Australian Fit: 2/5. ## Score | Criterion | Score | Why | | --------------------- | --------- | -------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | | Value for Money | 2/5 | Many models on one bill is real value, but credit burn, monthly expiry, off-brief re-generation and USD billing make true cost unpredictable and easily higher than the sticker. | | Ease of Use | 4/5 | Describe-a-brand-get-a-set is the easiest on-ramp in the category, and on-canvas layer editing lowers the skill floor further. | | Feature Depth | 4/5 | Agentic task planning, multi-model routing across Flux, GPT Image, Seedream and major video models, infinite canvas, brand kits and SVG export. Broad and ambitious. | | Australian Fit | 2/5 | No AUD, GST unstated, US data residency with no AU region, US operator and Chinese parent, no AU support. | | Reliability & Support | 3/5 | Capable and improving fast, but young, post-beta, with a public record of credit, lockout and support complaints and no local escalation path. | | **Total** | **15/25** | | ⭐⭐⭐ Good (11–15). Australian Fit is 2/5, below the 4/5 threshold in our [methodology](https://www.theaidirectory.com.au/methodology/), so Lovart cannot be named a category Top Pick regardless of its score — the raw 15/25 earns an honest three stars. ## Bottom line **If you want to ideate fast,** Lovart is exhilarating. For first-round brand directions, mood-setting and putting a spread of options in front of a client quickly, the agentic canvas does things single-image tools cannot, and the free tier lets you prove it costs nothing to find out. **If you are on a budget,** model the credits before you commit. One complex brand job can swallow a big chunk of a monthly allowance, credits expire, and an off-brief agent costs more to correct. Trial it on the free plan, run a real job, and watch the meter before you pick a paid tier — then buy non-expiring top-up packs rather than over-subscribing. **If your work holds Australian personal or client-confidential data,** the US-only data residency, the absence of an Australian region or privacy commitment, and the US-operator/Chinese-parent structure are the deciding issues. Treat data location as a requirement to be proven before you upload anything sensitive, not after. **If you are choosing between Lovart and a Canva-style tool,** they answer different questions. Lovart is for breadth and ideation — many models, full sets, one prompt. A tool like [Canva Magic Studio](https://www.theaidirectory.com.au/canva-magic-studio-review-australia-2026/) is for control, templates and team workflow with a far gentler cost story. Pick Lovart when you want the agent to surprise you; pick the safer suite when you need to know exactly what you will get and exactly what it will cost. Update history - **30 Sep 2026 — re-verified by browser; three of four paid prices had risen, and the free allowance is smaller.** `lovart.ai/pricing` on an Australian connection, monthly billing: **Starter US$19, Basic US$32, Pro US$90, Ultimate US$199** (the table had US$15, US$26, US$72 and US$199); billed annually they are US$16, US$27, US$45 and US$109, the last two on a promotion. Credits per plan hold. The free plan's daily refresh is about **30 credits**, not \~100 (paid plans get 100 a day). The note beside the table also said 10% GST had been added to figures that are ex-GST; corrected. Bare `$`, no `A$`. `aud_native: false` already set. Score unchanged. Capture: `Dossiers/captures/lovart-pricing-2026-09-30.txt`; [Wayback 30 Sep 2026](https://web.archive.org/web/20260929203439/https://www.lovart.ai/pricing). - **21 Sep 2026** — **Corrected a leftover from the retired exchange-rate basis.** This review's prose told the reader to expect the Australian dollar to add **“roughly 55% over the sticker”**, which is the fixed \~1.55 rate **our AUD pricing rule retired on 5 September 2026**. The figures in the pricing table were re-cut to the live rate by the FX sweep at the time; **the sentence describing them was not**, so the review has been contradicting its own footnote ever since — one line saying the A$ column was converted at 1.4045, another telling the reader to budget as though it were 1.55\. At 1.4045 the uplift is **40.45%**, so the sentence now says **about 40%** and cites the rate and its date inline, as our AUD pricing rule requires. **No price, figure or score changes** — the table was already correct; only the sentence a reader budgets from was wrong. - **20 Sep 2026** — Sora mention dated against OpenAI's sunset: the consumer app closed 26 April 2026 and the API closes 24 September 2026\. Score unchanged. - **25 Sep 2026** — Sora mention moved to the past tense: OpenAI's developer documentation confirms the Sora 2 models and Videos API were shut down on 24 September 2026. - **25 Jun 2026** — First reviewed. ## Frequently asked questions ### What makes Lovart a "design agent" rather than just an AI image generator? Lovart plans and executes the whole job, not one image. You give it a goal — say, a full brand identity — and it breaks the task into steps, runs them autonomously, and lays a coordinated set of logos, posters, social graphics and more across an infinite canvas. A generator gives you one output per prompt; an agent gives you a finished set. ### Is Lovart's pricing in Australian dollars? No. Pricing is in US dollars across a free plan and four paid tiers (Starter, Basic, Pro and Ultimate), all on a credit model. There is no AUD price and no GST line of Lovart's own — expect the exchange-rate premium plus likely GST through its payment processor, and confirm the final charge at checkout. ### How fast do Lovart's credits run out? Faster than most people expect. Because the agent runs many steps automatically, a single complex brand project can consume hundreds of credits, and credits reset monthly without rolling over. Heavier users report real spend climbing once top-up packs are added. Trial it on the free plan and watch the meter before committing to a tier. ### Can I use Lovart designs commercially in Australia? Yes. Lovart's terms state you retain ownership of your creations, and paid plans include commercial usage rights with no attribution required. As with any AI tool, you are responsible for ensuring your prompts do not infringe others' rights, and AI-generated logos sit in unsettled copyright territory — take legal advice before trademarking one. ### Is my data stored in Australia with Lovart? No. Lovart's privacy policy states data is stored on servers in the United States, with transfers to affiliates and providers in other countries. There is no Australian region, the operator is a US (Delaware) entity with a Chinese parent in LiblibAI, and the policy carries no Australia-specific commitment. Weigh that before uploading sensitive or client-confidential material. ### Which AI models does Lovart use? Several, routed by task. For stills it draws on Flux, GPT Image, Seedream, Nano Banana and Stable Diffusion; for video it includes Veo 3, Kling, Sora and Hailuo. The appeal is access to many leading models through one subscription, with Lovart picking — or you choosing — the right engine for each step. ## Read next - [Canva Magic Studio review: the everyday AI design suite, tested for Australia](https://www.theaidirectory.com.au/canva-magic-studio-review-australia-2026/) - [Recraft review: the vector-first AI design tool, tested for Australia](https://www.theaidirectory.com.au/recraft-review-australia-2026/) - [Best AI design and creative tools for Australians](https://www.theaidirectory.com.au/best-ai-design-creative-tools-australia-2026/) - [How we score AI tools — our review methodology](https://www.theaidirectory.com.au/methodology/) ## Sources - https://www.lovart.ai/pricing (primary; plans, credits, expiry, relax mode) - https://www.lovart.ai/statement/privacy-policy (primary; US data storage, operator Resonate International INC, international transfers) - https://www.lovart.ai/statement/terms-of-use (primary; ownership, commercial rights) - https://www.lovart.ai/features/all-in-one-ai-models (primary; model list — Flux, Veo 3, Kling, Nano Banana) - https://news.aibase.com/news/19935 (secondary; beta-to-general launch) - https://eu.36kr.com/en/p/3370076731868936 (secondary; 13 May 2025 beta, 20,000 waitlist) - https://pollo.ai/hub/lovart-ai-review (secondary; hands-on testing, agentic workflow, limitations) - https://www.trustpilot.com/review/www.lovart.ai (secondary; billing, lockout and support complaints) - https://magiclight.ai/academy/lovart-ai-review/ (secondary; credit-burn, real monthly spend) - https://help.apiyi.com/en/lovart-ai-expensive-alternatives-apiyi-guide-en.html (secondary; credit economics) Wayback Machine snapshots queued.