The Replicate alternative
Every model.
One API.
1,000+ fast, affordable production models with a listed price per run. No GPUs to pick, no idle instances to pay for. Try it here, then ship it from your code.
Get your API keyWhy developers switch from Replicate
Fast.
Blazing-fast inference on every model. Your images and videos come back fast, in the studio and through the API.
Affordable.
Images from $0.024, videos from $0.05 per run. Pay only for what you generate, with no subscription and no monthly fee.
- Hosted models
- 1,041+
- Model makers
- 43
- GPUs to choose
- 0
- Seconds of idle time billed
- 0
Know the price before you call
Every model has a listed price per output, right on its page. No hardware to pick, no per-second meter, no surprises on the invoice.
Replicate: Most models are billed by the time they take to run, at a per-second rate that depends on the hardware.
Zero idle time. Ever.
Every model is hosted and ready, so you pay only for the requests you send. Nothing to boot, nothing to keep warm.
Replicate: Private models run on dedicated hardware, billed for all the time instances are online, including setup and idle time.
Only production-grade models
A curated catalog of the leading closed and open models, from Seedance 2.5 to Nano Banana Pro, each with a schema, examples and a live price.
Replicate: Community and official models side by side, plus models you deploy yourself.
Studios for your whole team
Designers and marketers get image, video, audio and 3D generators, editing tools and an AI video editor in the browser, on the same account as your API.
Replicate: A web playground on each model page.
Popular models
Production endpoints, each with a schema, examples and a live price. Pick one and run it.

bytedance / seedance-2.5/image-to-video
Cinematic image to video, native audio included.

google / nano-banana-pro/text-to-image
Text to image that follows your prompt to the letter.

kwaivgi / kling-v3.0-pro/image-to-video
Buttery, controllable motion from a single still.

openai / gpt-image-2.5-sunburst/text-to-image
Precise images with crisp, legible in-image text.

alibaba / wan-3.0/image-to-video
Image to video with optional audio.

bytedance / seedream-v5.0-pro/edit
Reference-based image editing up to 2K.
elevenlabs / eleven-v3
Expressive text to speech.

meshy / v7.1/image-to-3d
Textured 3D models from one image.
How it works
Pick a model, send JSON, get a prediction id back instantly. Poll the result or let a webhook call you. That is it.
Run models
One request format for every image, video, audio and 3D model. Python and JavaScript SDKs, or plain HTTP. Learn it once, use it everywhere.
POSThttps://api.wavespeed.ai/api/v3/google/nano-banana-pro/text-to-image
curl -X POST "https://api.wavespeed.ai/api/v3/google/nano-banana-pro/text-to-image" \
-H "Authorization: Bearer $WAVESPEED_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"prompt": "A poolside patio at sunset with vintage lounge chairs",
"aspect_ratio": "4:3",
"resolution": "2k"
}'
# → {"data": {"id": "<prediction id>", ...}}
curl "https://api.wavespeed.ai/api/v3/predictions/<prediction id>/result" \
-H "Authorization: Bearer $WAVESPEED_API_KEY"import os, time, requests
headers = {"Authorization": f"Bearer {os.environ['WAVESPEED_API_KEY']}"}
task = requests.post("https://api.wavespeed.ai/api/v3/google/nano-banana-pro/text-to-image", headers=headers, json={
"prompt": "A poolside patio at sunset with vintage lounge chairs",
"aspect_ratio": "4:3",
"resolution": "2k"
}).json()["data"]
while True:
result = requests.get(f"https://api.wavespeed.ai/api/v3/predictions/{task['id']}/result", headers=headers).json()["data"]
if result["status"] == "completed":
print(result["outputs"])
break
if result["status"] == "failed":
raise RuntimeError(result.get("error"))
time.sleep(2)const headers = { Authorization: `Bearer ${process.env.WAVESPEED_API_KEY}`, 'Content-Type': 'application/json' }
const { data: task } = await fetch('https://api.wavespeed.ai/api/v3/google/nano-banana-pro/text-to-image', {
method: 'POST',
headers,
body: JSON.stringify({
"prompt": "A poolside patio at sunset with vintage lounge chairs",
"aspect_ratio": "4:3",
"resolution": "2k"
}),
}).then((r) => r.json())
while (true) {
const { data: result } = await fetch(`https://api.wavespeed.ai/api/v3/predictions/${task.id}/result`, { headers }).then((r) => r.json())
if (result.status === 'completed') { console.log(result.outputs); break }
if (result.status === 'failed') throw new Error(result.error)
await new Promise((r) => setTimeout(r, 2000))
}Made with WaveSpeedAI
/filters:quality(82)/examples/679e9087b2754a77991ae0e61fe4f9e8/a277a10d-3a03-48e0-9aa9-510db64d8bac-b1f7e25818a0.webp)
/filters:quality(82)/examples/9983000584e74cc581e2aa3596e0efc6/1783684219348367405_p8cYLApe.webp)
/filters:quality(82)/examples/b83e1ecb55e747809bf808dc76e9e323/e201e349-d4a3-4959-8873-1ae99847ca87-717a36cb81d5.webp)
WaveSpeedAI vs Replicate
| WaveSpeedAI | Replicate | |
|---|---|---|
| Catalog | Curated, production-ready image, video, audio, 3D and LLM endpoints | Community and official models, plus your own |
| Billing | A fixed price per run for each model, pay as you go | Pay for what you use: some models by hardware time, others by input and output |
| Custom models | No deployment needed; call hosted models directly | Deploy custom models and fine-tunes, billed for the time instances are online |
| Creative tools | Image, video, audio and 3D studios on the same account | Web playground on each model page |
| Async results | Poll the result endpoint or receive a webhook | Predictions with polling and webhooks |
Questions
Is WaveSpeedAI a good Replicate alternative?
If you want hosted, ready-to-call image, video, audio and 3D models with a clear price per run, yes. One API key, 1,000+ models, and you pay per request. That is the whole setup.
Do I need to choose hardware?
Never. Every model on WaveSpeedAI is a hosted endpoint with its own per-run price. No GPU to pick, no instance to size, no idle time to pay for.
How do I call a model?
POST JSON to https://api.wavespeed.ai/api/v3/<model> with your API key. A prediction id comes back instantly. Poll the result endpoint, or pass a webhook and we call you.
Is there a free trial?
Yes. Eligible new accounts get $1 in free credits to start calling models, with no credit card needed. Some premium models may not be available with trial credits.
Are there client libraries?
Yes. Install wavespeed for Python or JavaScript and you are a few lines from your first output. Any HTTP client works with the REST API too.
Run your first model today
$1 in free credits for eligible new accounts. No card needed.
Replicate is a trademark of its respective owner. WaveSpeedAI is not affiliated with, endorsed by or sponsored by Replicate. Comparisons reflect publicly available information at the time of writing and may change.
/filters:quality(82)/examples/883fa92158a04368bb87b6a2a2096351/1783684221888696489_duh4aXLA.webp)