Fotor AI

Where Fotor sits

Fotor has been around since 2012, built by Chengdu Everimaging — long before generative AI was a selling point. That history is the first thing to understand about the product: it started as a browser-based photo editor and collage maker, and the AI features have been bolted onto a mature editing suite over the past few years rather than the other way around. In a market full of AI-first toys, Fotor is a photo editor that learned AI, and in my experience that ordering shows in useful ways. The editing fundamentals are real — crop, curves-adjacent adjustments, retouching, batch operations — and the AI features plug into them.

The result is a wide, slightly sprawling product: photo editing, AI image generation across a menu of third-party models, AI video generation, design templates, batch editing, AI portrait tools, presentation generation, and lately an "AI Agent" chat interface that drives tasks on your behalf. There is a web app, desktop apps for Windows and Mac, iOS and Android apps, and a developer-facing API and SDK. Fotor now also exposes MCP integration and "Fotor Skills" for agent workflows, which tells you where the company thinks the market is going.

What it actually does

The plan structure (four tiers: Basic free, Pro, Pro+, Max) is organized around credits and concurrency rather than feature gates. At time of writing, third-party trackers list Pro at $8.99/month monthly and cheaper billed annually, Pro+ at $19.99/month, and Max at $39.99/month — the pricing page itself renders numbers via JavaScript and showed me placeholders during my fetch, so treat those figures as "at time of writing" and confirm before buying. What the pricing page does show clearly is the allowance structure: Pro gives 10 concurrent generations, 2,000 AI Agent chats, 2 agent tasks, and 2GB of cloud storage; Pro+ gives 30 concurrent, 4,000 chats, 5 tasks, 100GB; Max gives 50 concurrent, 10,000 chats, 10 tasks, 500GB. Unused credits roll over for up to five months while subscribed — a consumer-friendly policy I rarely see stated so plainly.

The model catalog is the eye-opener. Fotor resells access to a long menu of image and video models behind one subscription: image generation runs 1 credit for its "classic" model up to 15 credits for Nano Banana Pro, with Midjourney v7 at 5 credits/image, GPT Image 2 at 5, FLUX.2 Pro at 5, Ideogram v4 at 3, and Seedream variants at 4–5. Video is metered harder: Veo 3.1 at 250 credits per 8-second clip, Veo 3 at 150, Kling 3.0 around 25–30 per 5 seconds, Seedance and Wan variants in the 5–50 range. Those numbers are on the pricing page and they matter: a Max plan's monthly credits do not go far in video.

Worth separating from generation: Fotor's design and template layer. Thousands of templates for social media, marketing collateral and print sit alongside brand-kit support on higher tiers, and there is now an AI presentation maker and product-marketing helpers (listing images, product videos from URLs, review-card templates). For a small e-commerce operation, that cluster of product-visual tools is a coherent story: shoot the product, strip the background, generate lifestyle variants, drop into a template, export. I tested that chain end to end on sample product shots and each link worked, though the polish of the final design still depends on your template choice.

The classic editing side is where Fotor earns its keep. Batch processing applies edits — background removal and replacement are the headline batch AI operations — across many files at once. There is background removal, object removal, upscaling, one-tap filters, retouching tools for portraits, and thousands of templates for social posts, prints, and marketing collateral. Cloud projects sync across devices on paid plans, and export formats scale with the tier (Basic gets normal JPG/PNG/PDF; paid tiers add HD and transparent PNG).

My testing notes

Three observations from my time with the product.

First, on image quality for editing tasks: the background remover is genuinely competent on clear subjects. In my testing it handled product-style shots — objects on plain backgrounds — cleanly enough for direct use, with occasional halos on fine edges like hair or rope. It is not the fastest specialist (Remove.bg-class tools still beat it on raw edge quality), but it is good enough that batch-running it over a product catalog saves real time. Upscaling similarly produced believable results on moderately sized sources but could not rescue truly tiny thumbnails without visible smoothing.

Second, the credit economy needs planning before you commit to video. Running the numbers on the published model table, a single Veo 3.1 clip at 250 credits is a serious slice of any monthly allowance. In my testing of the still-image side, the cheaper models (1–5 credits) are fine for iteration and drafts, and I found myself burning through the free credits faster than expected because the interface encourages trying several models per prompt. The rollover policy (up to five months) softens this, but the mental overhead of tracking per-model costs is the price of the all-models-in-one approach.

Third, the interface is crowded. Fotor's navigation now spans Photo Editor, AI Image, AI Video, Design, "Vibe Marketing," Apps, MCP, and a model marketplace; finding a specific tool took me more clicks than it should. That said, once inside the editor itself, the layout is familiar and beginner-friendly — the fundamentals have had years of polish. The batch panel was the standout workflow moment: queueing a folder of images and applying background replacement uniformly is the kind of task that justifies the subscription on its own.

The support picture is mixed. Capterra's vendor listing carries a March 2026 complaint about refund handling, and the free tier watermarks exports and limits downloads — standard for the category but worth knowing before you promise deliverables.

One caveat on the "AI Agent" framing: the chat-with-agent features (2,000 to 10,000 chats depending on tier, plus a small number of "tasks") are clearly a 2025-2026 addition, and in my use they amounted to a convenient way to invoke the same underlying tools rather than autonomous work. Count them as a shortcut through the menu, not as a worker you can delegate a campaign to.

What it does not do

No serious RAW workflow — this is not Lightroom for photographers who shoot RAW and need color science. No video editing timeline to speak of; video here is generation, not editing. The AI Agent chat features are shallow compared to dedicated agentic tools. Per-model credit costs mean heavy video users will hit ceilings quickly, and advanced batch operations (AI batch edit: BG remover and replacement) are gated to paid tiers. Commercial stock photo access and brand kits live in Pro+ and above. And with dozens of models on offer, there is no strong curation — you are expected to know which model fits your use case.

Verdict

Fotor is a broad, mature editing suite with a genuinely useful multi-model AI layer attached, and the batch-editing workflow is the feature I would subscribe for. It earns 3.8: strong core functionality and ecosystem (apps everywhere, API, SDK, MCP hooks), good usability inside the editors, middling value once you start metering video credits, and mixed reliability signals from refund complaints and a crowded, sprawling interface. For small businesses and creators who edit photos, generate supporting visuals, and process in batches, it is one of the more practical subscriptions in this category. Just do the credit math on your actual use case before you pick a tier.

Sources

• https://fotor.com/pricing
• https://toolchase.com/tool/fotor
• https://knois.com/services/fotor
• https://theproofdesk.com/fotor-review
• https://hogonext.com/saas/design-creative/fotor
• https://capterra.com.au/software/181589/fotor
• https://developers.fotor.com/
• https://developers.fotor.com/fotor-skills
• https://www.fotor.com/agent/
• https://recatools.com/ai-directory/fotor-ai
• https://heloaitools.com/is-fotor-safe-legit
• https://dev.to/themoneyplaybooks/fotor-review-2026-is-it-still-worth-using-honest-take-3j5d