Krea is a creative platform known for its real-time generation canvas, image enhancer and its own image models. Krea 2 Large is the company's text-to-image model, tuned for a photographic look that avoids the over-smoothed, plastic skin and saturated palette common in diffusion output. This repository is a small Python client for the Krea API as hosted on Synexa, so you can generate Krea 2 Large images from a script with one pip install and an API token.
You get a blocking run() that takes a prompt and returns image URLs, a non-blocking create-and-poll path for batch jobs, and webhook delivery for services that would rather be called back. The client has a single runtime dependency and ships no weights. It is aimed at developers adding image generation to a product, a content pipeline or an internal tool who want Krea's model as an HTTP call rather than a GPU deployment.
Try it now: https://synexa.ai/explore/krea/krea-2-large — the hosted model behind this client. New accounts get a free trial credit.
- Why this client
- Installation
- Quickstart
- Hosted models
- Parameters
- Advanced usage
- About Krea
- Use cases
- FAQ
- License
- The weights are not published. Krea 2 Large is a hosted model; there is no checkpoint to download, so a hosted endpoint is the only way to call it programmatically outside Krea's own app.
- No GPU or serving stack. Large text-to-image models of this class need a datacenter GPU with tens of gigabytes of VRAM and a serving layer with batching and queueing. The endpoint runs on Synexa's fleet and you pay per image.
- No cold start. The model stays resident on the endpoint; the first request of the day costs the same as the hundredth, and there is nothing to warm up on your side.
- Predictable cost.
krea/krea-2-largeis billed at $0.06 per run, so a thousand-image batch costs $60 and you know that before you submit it.
pip install git+https://github.com/krea-dev/krea-api.gitThen set your API key (create one at synexa.ai):
export SYNEXA_API_KEY="sk-..."import krea_api
output = krea_api.run({
"prompt": "A cinematic shot of a lighthouse at dawn, soft fog, warm light"
})
print(output) # URL(s) of the generated resultOr with an explicit client:
from krea_api import Client
client = Client(api_key="sk-...")
output = client.run({"prompt": "A cinematic shot of a lighthouse at dawn, soft fog, warm light"})| Model | Category | What it does | Price / run |
|---|---|---|---|
krea/krea-2-large |
text-to-image | Krea 2 Large generates high-fidelity images from text with a distinctly photographic, non-plasticky look. | $0.06 |
The default model is krea/krea-2-large; pass model="owner/name" to run() to use another one from the table.
| Field | Type | Required | Default | Range | Description |
|---|---|---|---|---|---|
prompt |
string | yes | A rain-slicked Tokyo side street at nigh… |
— | Text description of the image to generate. |
aspect_ratio |
string | no | 1:1 |
1:1, 4:3, 3:2, 16:9, 2.35:1, 4:5, 2:3, 9:16 | Aspect ratio of the generated image. |
creativity |
string | no | medium |
raw, low, medium, high | Controls how loosely the model interprets the prompt. Higher values produce more creative results that may drift from the prompt; lower values stay closer to the prompt. |
seed |
integer | no | random |
— | Random seed for reproducible generation. |
Submit without blocking, then poll:
prediction = client.run(input, wait=False) # returns immediately
prediction = client.wait(prediction, timeout=300)
print(prediction["output"])Webhook on completion:
client.run(input, wait=False, webhook="https://your-app.example/hooks/synexa")Errors:
from krea_api import ModelError, PredictionTimeout
try:
output = client.run(input)
except ModelError as e:
print("failed:", e, e.prediction and e.prediction.get("id"))
except PredictionTimeout:
print("still running — poll later")Status values you will see on a prediction: starting → processing → succeeded | failed.
Krea (krea.ai) is a creative AI platform. It became known for a real-time canvas where an image updates as you draw or move shapes, and has since added an upscaler and enhancer, video generation, training on your own images, and a model hub that fronts third-party image and video models alongside its own. Krea also collaborated with Black Forest Labs on FLUX.1 Krea [dev], an open-weights model that targets the same anti-plastic aesthetic.
Krea 2 Large is Krea's own text-to-image model. Its stated goal is realism without the tells that make AI images recognisable: waxy skin, uniform bokeh, over-contrasted lighting and a narrow colour range. Given a text prompt it produces a single high-resolution image; the hosted endpoint exposes an aspect_ratio control, a creativity slider that decides how loosely the prompt is interpreted (higher values drift further from the literal prompt, lower values stay close to it) and a seed for reproducible output. The model is strongest on photographic subjects such as people, products, interiors, landscapes and editorial scenes, and is not a specialist for typography or diagrams.
Typical usage is product and lifestyle imagery, stock-style photography for landing pages and ads, concept art with a photographic finish, and seed-locked variations where you want to change the prompt slightly while holding composition. Output is one image per run; batch by issuing several runs, which the poll-based path handles without blocking.
The hosted endpoint used by this client is krea/krea-2-large on Synexa, which is Krea's own Krea 2 Large model served through Synexa's API. It is not a substitute or look-alike; the same model is available in Krea's app and on Krea's own platform at krea.ai, which also offers features this client does not cover, such as the real-time canvas, enhancer and custom training.
Official project: https://www.krea.ai
- Product and lifestyle imagery — call
run()with a prompt describing the product, setting and lighting, and anaspect_ratiomatching the ad slot. - Landing-page hero images — generate a set of candidates from one prompt with different seeds through the poll path and pick the best in review.
- Editorial illustration — describe the scene and set
creativitylow to keep the model on the brief for article headers that must match the copy. - Concept exploration — raise
creativityand vary the prompt to get looser interpretations early in a design process. - Seed-locked variations — fix
seed, change one phrase in the prompt, and compare how the composition holds across runs. - Programmatic asset pipelines — enqueue hundreds of prompts from a spreadsheet and receive results by webhook as they complete.
Is there a Krea API?
Yes. Krea offers an API on its own platform, and Krea 2 Large is also hosted on Synexa as krea/krea-2-large. This client talks to the Synexa endpoint, which takes a text prompt and returns a generated image.
How much does the Krea API cost through this client?
The hosted krea/krea-2-large endpoint is billed at $0.06 per run, where one run produces one image. There is no subscription; you pay per completed generation.
Can I run Krea without a GPU?
With this client, yes. Generation happens on Synexa's GPUs; your machine only needs Python and network access. Krea 2 Large weights are not published, so there is no local option to compare against.
Does this client work with ComfyUI or FLUX.1 Krea [dev]?
No. It does not load local checkpoints or drive ComfyUI; it only calls the hosted endpoint. FLUX.1 Krea [dev] is a separate open-weights model and is not what this endpoint serves.
What input formats does it accept?
The only required field is prompt, a text description of the image. Optional fields are aspect_ratio, creativity (how loosely the prompt is interpreted) and seed (for reproducible output). There is no image input on this endpoint.
Is this the official Krea SDK?
No. This is an independent client that wraps the Synexa-hosted endpoint. Krea's official product and API are at https://www.krea.ai.
- Krea — official platform, app and API
- Synexa Python client — the general-purpose SDK this client builds on
- krea/krea-2-large on Synexa — the hosted Krea 2 Large endpoint
- black-forest-labs/flux-schnell on Synexa — a fast, low-cost alternative for drafts
- google/nano-banana-pro on Synexa — image generation and editing with reference images
MIT. This is an independent, community-maintained client and is not affiliated with or endorsed by the authors of Krea. Model weights and trademarks belong to their respective owners.