How much does WaveSpeedAI cost?

WaveSpeedAI charges per run from a prepaid balance, with no plan and no monthly fee. New accounts get $1 of credit with no card. Nano Banana 2 is $0.07 an image at 1K, and Veo 3.1 Fast is $0.15 a second with audio.

Is WaveSpeedAI cheaper than Visual Sandbox?

On Google models such as Nano Banana 2 and Veo 3.1 Fast, yes, by about our markup. On Seedance 2.0 Fast from text or an image, Visual Sandbox costs less. Neither has a plan, and a month you make nothing costs $0 on both.

Does WaveSpeedAI run the models itself?

It runs its own serving software: an in-house compiler, GPU kernels and a scheduler. That speed-up applies to the open models it hosts, such as Wan and FLUX. Closed models such as Nano Banana and Veo run on Google's servers, for every host.

Does Visual Sandbox have a CLI and an MCP server?

Yes. The CLI, the MCP server and the API spend the same balance as the web app, at the same price.

All comparisons

Visual Sandbox vs WaveSpeedAI

Both charge per run. Both talk to agents. Only one has a canvas.

In short

WaveSpeedAI charges per run. No plan, no monthly fee, $1 free to start.
On Google models it is cheaper than us, by about our markup.
On Seedance 2.0 Fast from text or an image, we are cheaper.
Both ship a CLI, an MCP server and an API. Only we give the agent a canvas.

Our take. Use WaveSpeedAI for the lowest price per API call. Use Visual Sandbox when you and your agent make the work together.

See a model's price

WaveSpeedAI

WaveSpeedAI hosts 1000+ image, video, audio, 3D and LLM models behind one API, sold per run with $1 free, and reached from a CLI, an MCP server and agent plugins.

Pricing: Per run, from a prepaid balance
Free plan: $1 of credit, no card
Also does: 1000+ models, LLMs, a CLI and an MCP server

Side by side

The same 9 questions, asked of WaveSpeedAI

Visual Sandbox compared with WaveSpeedAI, question by question
What is compared	Visual Sandbox	WaveSpeedAI
What you pay for	Each generation, at our listed price	Each generation, at their listed price
What the balance is in	Money. One credit is one cent	Money. A prepaid dollar balance
Free tier	None. A $10 minimum top up	$1 of credit on signup, no card
Cheapest paid tier	There is no plan	There is no plan
A month you make nothing	$0	$0
The price before you run	A dollar figure on every model page, before you sign in	A dollar figure on every model page, and a quote from the MCP server
Working as a team	Not offered	Not on the pricing page
From the terminal or an agent	A CLI, an MCP server and an API, on the same balance	A CLI, an MCP server, an API and plugins for many agents
The catalogue	90 models across image, video, audio, 3D and vector	1000+ models across image, video, audio, 3D and LLMs

Key takeaways

Is Visual Sandbox a good WaveSpeedAI alternative?

Where we win

A canvas you both see: Runs land on a canvas. Click an image and the agent knows which one you mean.
Skill packs for the agent: The MCP server hands an agent our prompting guidance before it prompts. Their skill teaches the call, not the prompt.
Free video edits, locally: The CLI cuts, crops, retimes and joins clips on your machine, at no cost.
Cheaper on Seedance 2.0 Fast: Seedance 2.0 Fast is $0.09 a second at 480p here, from text or an image. WaveSpeedAI charges $0.10.
Presets you can share: Save a model and its inputs as a preset with a public link.

Where WaveSpeedAI wins

1000+ models: Image, video, audio, 3D and LLMs on one balance.
Lower prices on Google models: Nano Banana 2 at 1K is $0.07. Veo 3.1 Fast with audio is $0.15 a second.
Plugins for many agents: Claude Code, Codex, Gemini CLI, Copilot CLI, Kimi and Zed, plus nodes for ComfyUI, n8n and Dify.
Their own speed-up: An in-house compiler and GPU kernels make the open models they host, such as Wan and FLUX, run faster.
A dollar to start: $1 of credit, no card.

Worth checking

WaveSpeedAI is cheaper per run: On the Google models, by about our markup. That is the honest headline.
Not on every model: From text or an image, Seedance 2.0 Fast costs less here, at 480p and at 720p.
Listed prices can move: Their model pages call a price a reference. The charge is set when the run starts.
The API comes first: SDKs, a CLI and an MCP server lead. A web generator and a desktop app are there for everyone else.
No canvas for the agent: Their CLI and MCP server return URLs. Nothing tells the agent which result you picked.

Two ways to pay per run

The price is on the page before the first run

This is the live price card off our Seedance 2.0 Fast page. WaveSpeedAI lists the same model at $0.10 a second at 480p and $0.20 at 720p.

No account needed to read it, on either site.
On Nano Banana 2 and Veo 3.1 Fast, their price is lower than ours.
A quiet month costs $0 on both.

WaveSpeedAI sells the call. We sell the call and the canvas it lands on.

Seedance 2.0 Fast

What one run costs, before any account

480p: $0.09/s
480p: $0.18/s
720p: $0.19/s
720p: $0.38/s

See it on the model page

The money

Where the money goes

Both charge per run and neither bills a plan. The gap moves model by model.

$0.07

WaveSpeedAI

Nano Banana 2 at 1K.

$0.09

Visual Sandbox

The same image, with our markup and the canvas around it.

WaveSpeedAI, to start

Free credit, no card. Our minimum top up is $10.

Row	Visual Sandbox	WaveSpeedAI
Nano Banana 2, 1K	$0.09	$0.07
Nano Banana 2, 4K	$0.19	$0.14
Veo 3.1 Fast with audio, a second	$0.19	$0.15
Seedance 2.0 Fast 480p, from text, a second	$0.09	$0.10
Seedance 2.0 Fast 720p, from text, a second	$0.19	$0.20
FLUX.2 Klein 9B, an image	$0.02 at 1 MP	$0.01 at any size
Plan or fee	None	None
To start	$10 minimum top up	$1 free, no card
A canvas the agent can read	Yes	No

Nano Banana and Veo are closed models. Every host, WaveSpeedAI and us included, runs them through Google. The gap on those is margin, not speed.
Their speed-up applies to the open models they host, such as Wan, FLUX and Z-Image. That is where a GPU host has a real cost edge.
If you only need the cheapest API call, WaveSpeedAI wins on Google models. If you want the canvas and the prompting guidance, the markup here pays for them.

Which one fits

Which one fits you?

Four questions. The answer names WaveSpeedAI when WaveSpeedAI is the right call.

01 of 04

Are you building a product on top of the models?

Or use both

We cover the making. WaveSpeedAI covers the volume, if you ship software.

01
Try models and make the assets here, on a canvas your agent can see.
02
Send your own product's calls to WaveSpeedAI's API at their rate.
03
Keep Seedance 2.0 Fast runs from text or an image here, where they cost less.

The call