Back to guides
Image·July 2, 2026·6 min read

Seedream API (incl. Seedream 4.5) — what it is, pricing, and the live alternative

Seedream is ByteDance's text-to-image family, known for character consistency. Here's what the Seedream and Seedream 4.5 APIs do, how they're priced, and how to run the same workflow on Kunavo today with Nano Banana or GPT-Image-2.

Seedream is ByteDance's text-to-image model family — Seedream v4, Seedream 4.5 and Seedream 5 Lite — known for strong character consistency across generations. The Seedream API generates and edits images from a text prompt programmatically. This guide explains the family, its pricing, and how to run the same workflow on Kunavo today.

Availability note. Seedream is on Kunavo's roadmap but not currently enabled. ByteDance's own price is already very low (~$0.03–0.04 per image), and Kunavo only lists a model when it can serve it below the provider's direct rate — so for now Seedream isn't offered. The live text-to-image models are Nano Banana and GPT-Image-2 on the same OpenAI-style endpoint; switching to Seedream later is a one-word change to the model field.

The Seedream model family

ModelByteDance list (approx.)Known for
Seedream v4~$0.03 / imageCharacter consistency, flagship quality
Seedream 4.5~$0.04 / imageRefined v4, better prompt adherence
Seedream 5 Lite~$0.035 / imageFast, cost-effective

Because ByteDance prices Seedream so aggressively, going direct is currently the cheapest route — which is exactly why Kunavo hasn't enabled it. We'd rather tell you that than resell it at a markup.

Text-to-image on Kunavo today

Kunavo exposes image generation through one OpenAI-style endpoint, POST /v1/images/generations. The live models are Nano Banana and GPT-Image-2. The request shape is identical to what a Seedream call would be — only the model string differs:

text_to_image.py
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["KUNAVO_API_KEY"],
    base_url="https://api.kunavo.com/v1",
)

# Live text-to-image on the same OpenAI-style endpoint. Swap the model
# string to Seedream once it's enabled — the request shape is identical.
img = client.images.generate(
    model="nano-banana",   # or nano-banana-pro / gpt-image-2
    prompt="a character sheet of a fox ranger, consistent face, 3 poses",
    size="1024x1024",
)
print(img.data[0].url)

The closest live alternatives

  • Nano Banana — Google's image family, strong character consistency, about $0.027 per image. The nearest match to Seedream's strengths. See the Nano Banana API guide.
  • GPT-Image-2 — best prompt fidelity and text-in-image, about $0.0633 per image. See the GPT-Image-2 API guide.
  • For image editing, both offer image-to-image via /v1/images/edits — details in the images reference.

FAQ

What is the Seedream API?

ByteDance's text-to-image model family (v4, 4.5, 5 Lite), known for character consistency, priced around $0.03–0.04 per image on ByteDance's list.

Can I use Seedream on Kunavo today?

Not yet — it's on the roadmap. ByteDance's direct price is already so low that Kunavo can't beat it, and we only list models we can serve cheaper. Use Nano Banana or GPT-Image-2 on the same endpoint today.

What is Seedream 4.5?

A refined Seedream v4 with better prompt adherence, around $0.04 per image. On Kunavo, the live equivalents are Nano Banana and GPT-Image-2.

What's the best Seedream alternative?

Nano Banana for consistency at low cost; GPT-Image-2 for prompt fidelity — both live on Kunavo through the OpenAI-compatible images endpoint.