Kinovi pricing, free plan and limits

Kinovi is a pay-as-you-go API and browser studio that runs dozens of video, image and audio models from ByteDance, Alibaba, xAI and others on one credit balance.

Best for
Independent filmmakers and directors creating previz and pitch reels under tight budgets
Not for
Casual social media creators wanting one-click video generation without learning @-tags

Should you use Kinovi?

Kinovi is a credit-based API and browser studio that runs 43 video, image and audio models from ByteDance, Alibaba, xAI, MiniMax, OpenAI and Suno. Credits are one-time, never expire and cover both surfaces. It suits developers who want cheaper Seedance calls than Fal.ai lists, but purchases are non-refundable.

The ability to pin First & Last Frame while the AI generates the middle, combined with @-tag shot direction across 9 image and 3 video references, offers unprecedented control for previz and commercial work compared to one-button prompt generators.

Kinovi's reference-driven workflow and @-tag direction are novel UX choices, but the underlying Seedance 2.0 model is not proprietary. Competitors can replicate the directing UX quickly. Long-term moat depends on building a director-centric community and staying ahead on UI/feature parity with Runway, Veo, and newer models.

Also for

  • Creative agency producers generating product spots and commercial content with precise direction
  • Developers who want ByteDance, Alibaba and xAI video models behind one REST API
Show 3 more
  • Agent builders who want to price and submit generation jobs through MCP
  • Creators who prefer non-expiring credits to a monthly video subscription
  • Small studios comparing per-second video rates across resellers

Also not for

  • Large content teams needing unlimited concurrent generations and long-form video output

Main trade-off

4-15 second clip limit forces multi-shot storyboarding for anything longer, and lack of disclosed founding/funding raises sustainability concerns for a user-dependent tool.

Kinovi pricing

PlanMonthly priceWhat it includes
Credit packagesCustom43 video, image and audio routes, REST API and remote MCP server, Failed tasks refunded automatically, Concurrency starts at 3 tasks
EnterpriseCustomPriority onboarding, Custom invoicing, Higher concurrency
How the pricing works, in full

Kinovi sells one-time credit packages with no subscription; credits never expire and one balance covers the Studio and the API. Its public pricing page (read 12 November 2026) lists per-model rates rather than package prices: the cheapest video route costs $0.0125 per second of output, the newest Seedance model $0.1197, and GPT Image 2 $0.0100 per image. Some regions give new accounts a small starter grant. Purchases are non-refundable except for duplicate charges or non-delivery; failed tasks are refunded automatically. Enterprise volume and invoicing are by quote.

Before you commit

  • Concurrent generation limits (2 on Starter, 5 on XL) force queue management for production teams; more concurrent slots require higher-tier plans
  • 4K support only on XL Pack; Starter is limited to 2K, artificially segmenting quality by plan
  • Credit costs for Fast vs Pro models vary 2x, and per-second rates ($0.20-0.40) are opaque versus transparent pricing

What Kinovi does

Kinovi is not free beyond a small starter grant that new accounts get in some regions. It sells one-time credit packages, and its pricing page on 12 November 2026 listed 43 models; the cheapest video route cost $0.0125 per second and the newest Seedance model $0.1197.

  • Video, image and audio on one balance

    One key and one credit balance reach video routes from ByteDance, Alibaba, xAI and MiniMax, image routes from OpenAI and Midjourney, and music from Suno.

  • Public REST API

    createTask submits a model id and its inputs, recordInfo returns status, output URLs and credits used, and an optional callback URL receives one POST when a task ends.

  • Remote MCP server

    An MCP endpoint at kinovi.ai/api/mcp covers model discovery, cost estimates, task submission, polling and balance checks, with OAuth 2.1 sign-in.

  • Published per-model rates

    Each model page states its rate before you start, such as 4.5 credits a second ($0.0212) for the Seedance Mini route.

Show all 6 features
  • Automatic refunds on failure

    Credits are reserved when a task starts and returned if the provider fails, times out or blocks the input, so a task with no output costs nothing.

  • Browser Studio

    The same models run in a web Studio with characters, templates and a gallery, drawing on the same credit balance as the API.

Works with

  • API access (Starter and above)
  • Seedance 2.0 Pro and Fast models

Missing integrations

  • No Zapier or Make support
  • No Slack, Discord, or webhook integrations
  • No Premiere Pro or DaVinci Resolve plugins
  • No batch processing via FTP or cloud storage

Most users expect these and they are not currently supported.

Capabilities

  • Model API
  • Video generation
  • Image generation
  • Audio generation
  • Seedance
  • MCP
  • Pay as you go

Strengths and limits

Strengths

  • Seedance rates sit well below Fal.ai's on Kinovi's own comparison table, 46% lower on its newest Seedance model.
  • A single API key covers several model makers at once, which saves signing up for and wiring together separate vendor accounts.
  • Credits never expire and failed tasks are refunded automatically, so bursty projects lose nothing to unused balances or provider errors.
  • Developer docs, an OpenAPI file and a remote MCP server mean an agent can price and submit jobs without a custom integration.

Limits

  • The public pricing page lists per-model rates but no package prices, so the cost of the smallest top-up is only visible at checkout.
  • Credit purchases are final; refunds cover only duplicate charges or non-delivery.
  • Retention differs by surface: the API policy keeps generated media for 30 days, while the Studio privacy policy says 7 days, so download anything you need.
  • The privacy policy is silent on whether Kinovi trains on what you upload or type.
Show 2 more
  • New accounts start at 3 concurrent tasks, and higher tiers are assigned by Kinovi rather than bought.
  • A reseller sits between you and the model maker: a model's availability and terms depend on both.

How Hokai rates it

  • Ease of use7.0 / 10
  • Features8.0 / 10
  • Value for money6.0 / 10
  • Support6.0 / 10

Kinovi alternatives

ToolPrice fromFree planRuns onRating
Kinovi This pageQuoteNoWeb3.4
ReplicateReplicate bills GPU time across thousands of open models; Kinovi bills per second or per image for a curated set of commercial video, image and audio models.FreeYesWeb--
HiggsfieldHiggsfield sells monthly credit plans to creators with a studio first; Kinovi sells one-time credits and leads with an API and MCP server.$9/moYesWeb4.0
RunwayRunway pairs its own video models with an editor on a subscription; Kinovi resells other makers' models on pay-as-you-go credits.$15/moYesWeb · iOS · Android4.5

Kinovi facts

Data retention
30 days
Works offline
No
Minimum skill level
Intermediate
Price band
Mid-range
Headquarters
Hong Kong (Vixion Technology Limited)
Training-data policy
Not addressed: the privacy policy read on 12 November 2026 does not say whether uploads or prompts are used for model training.

Data and privacy

Proprietary platform under Kinovi's terms; users keep the rights to what they upload and generate (privacy policy, read 12 Nov 2026)

Questions about Kinovi

How much does Kinovi cost?

Kinovi charges per model from one-time credit packages, with no subscription. Each model has its own rate: on 12 November 2026 the pricing page put Seedance 2.5 at $0.1197 per second of video, the Lite version of Grok Imagine Video at $0.0125, and GPT Image 2 at $0.0100 an image. Package totals appear at checkout.

Is Kinovi free?

Not as a plan. Its FAQ says that, depending on region, new accounts receive a small starter grant to try the models. After that every generation draws on purchased credits, which never expire.

Does Kinovi have an API?

Yes. A REST API takes a public model id through createTask and returns status, output URLs and credits used through recordInfo, with an optional webhook. A remote MCP server at kinovi.ai/api/mcp offers the same jobs to AI agents, with OAuth sign-in.

Can I get a refund on Kinovi credits?

Generally no: Kinovi's pricing FAQ says credit purchases are final, with exceptions only for duplicate charges or non-delivery under its refund policy. Tasks that fail are different, and their reserved credits return to the balance automatically.

How long does Kinovi keep my videos?

Its API data retention policy deletes generated media 30 days after generation and uploaded media and request logs after 2 months. The Studio privacy policy says created content is stored for 7 days, so download anything you want to keep.

About Kinovi

Kinovi launched as a browser studio built around ByteDance's Seedance video model, sold in one-time credit packs. By November 2026 it describes itself as "one API for video, image, and audio models": a public REST API and a remote MCP server that run Seedance 2.5, Wan 3.0, Grok Imagine, MiniMax H3, GPT Image 2, Midjourney and Suno routes, with the browser Studio kept as an optional front end on the same balance. It is operated by Vixion Technology Limited, a company registered in Hong Kong, and says it is not affiliated with the model makers whose models it runs.

Read the full background

On 12 November 2026 its public pricing page listed 43 routes: 27 video, 8 image and 8 audio, each with a per-unit rate. Video ran from $0.0125 a second for Grok Imagine Video 1.5 Lite to $0.1197 a second for Seedance 2.5, and the page set its Seedance rates beside Fal.ai's, 46% below them for Seedance 2.5. Credits are one-time packages with no subscription, they do not expire, and a task that fails is refunded to the balance automatically. A developer submits a job with createTask and polls recordInfo, which reports the output URLs and the credits used.

The pick depends on what you need beyond the model call. Replicate bills open models by the second of GPU time, roughly $0.36 to $20 an hour depending on the hardware, per its pricing page; Higgsfield sells creator plans by the month, from $9 for 120 credits on its pricing page of 24 September 2026; and Runway or Kling suit teams who want one vendor's own model and editor. Kinovi is the narrower choice: a buyer who wants ByteDance's video models at a lower per-second rate than Fal.ai and a single key for video, image and audio. Who should not buy it: anyone who needs a refund path, since credit purchases are final, or a published package price before signing up, since the public page lists per-model rates but no package prices. The Seedance reseller route is compared in where to use Seedance for free, and the studio alternatives in best Higgsfield alternatives.

Sources and guides

More AI Tools on Hokai

  • Tamarind BioHosted platform for AlphaFold, RFdiffusion, Boltz and hundreds of other biology tools via web app, API and MCP, with a Free plan to start on.
  • ArchilA durable shared disk for AI agents that sits on top of your object storage, with serverless sandboxes that bill per tool call and a free tier.
  • Google AI StudioFree web-based IDE for building, testing, and deploying generative AI applications with Google's Gemini models
  • GroqFast open-model inference, now built on NVIDIA-licensed LPU hardware
  • Fireworks AIEnterprise inference platform for open models: serverless tokens from $0.10 per 1M, managed fine-tuning per training token, and dedicated NVIDIA GPUs from $8.00/hr. $1 free credits, self-serve, no cold starts.
  • Cast.aiCast.ai cuts Kubernetes cloud costs 50-70% using automated rightsizing and Spot instance management. Trusted by 2,100+ companies, reached $1B unicorn in 2026.