By No Code MBA · Facts checked
- Provider
- Tencent
- Announcement date
- Status at last check
- Limited access or preview
- Model family
- Hy Image
- Preceding model
- Hy Image 3.0
- Capabilities
- Image
What changed
Hy Image 3.5 Preview is Tencent Hunyuan's next image generation model, supporting text-to-image, single and multi-image reference generation, and multi-turn conversational editing. Tencent says it adds, removes, modifies and expands image content precisely while keeping characters and scenes consistent, aimed at storyboard and character assets. The API defaults to 1.5K output with a 2K tier, and accepts an explicit size up to 4096 by 4096 for 4K output.
Access and availability
Available through Tencent Cloud TokenHub as hy-image-v3.5-preview at the international endpoint tokenhub-intl.tencentcloudmaas.com, after enabling postpaid billing for the model in the console. Requests use a messages format, and multi-turn edits pass back the previous response's history.
Official access documentationPricing
$1.60 per million tokens, about $0.024 per image at 1K or 2K (15,000 tokens) and about $0.032 per image at 4K (20,000 tokens).
Official pricingHow it compares
Hy Image 3.5 Preview is the cheapest of these at 2K and 4K, and it handles reference images and multi-turn edits in one API. Pick Nano Banana 2.1 if you already build on the Gemini API, or Seedream 5.0 Pro if you want ByteDance's editing controls and can pay more per image.
| Model | Price | Best for | Source |
|---|---|---|---|
| Hy Image 3.5 PreviewTencent · This model | About $0.024 per 2K image ($0.032 per 4K image) | Low cost generation and multi-turn editing | Official page |
| Hy Image 3.0Tencent · Previous model | About $0.032 per image (one rate at every size) | Existing Hunyuan image workflows | Official page |
| Nano Banana 2.1Google · Competitor | About $0.050 per 2K image ($0.113 per 4K image) | Gemini API image apps | Official page |
| Seedream 5.0 ProByteDance (BytePlus) · Competitor | About $0.09 per 2K image ($0.045 at 1.5K or below) | Controllable generation and editing | Official page |
Prices and limits are from each provider's official page as of October 27, 2026.
Where you can use it
- Tencent Cloud TokenHubCall model hy-image-v3.5-preview on the international endpoint after enabling postpaid billing for it in the console.Source
Call it from code
import os
import requests
resp = requests.post(
"https://tokenhub-intl.tencentcloudmaas.com/v1/wand/hunyuan-image/v35-generation",
headers={"Authorization": f"Bearer {os.environ['TOKENHUB_API_KEY']}"},
json={
"model": "hy-image-v3.5-preview",
"size": "2048x2048",
"messages": [{
"role": "user",
"content": [{"type": "text", "text": "A golden retriever running on the beach at sunset"}],
}],
},
)
data = resp.json()
print(data["choices"][0]["delta"]["image"]["url"]) # URL expires after 12 hoursWhat you can build
Integrate Hy Image 3.5 Preview's messages-based API for iterative image editing workflows where users refine visuals through conversation rather than one-shot generation, useful for e-commerce and content production pipelines.
Limitations
The model is labeled preview. The API documents a limit of 20 reference images per conversation, and older turns are truncated beyond that. API regions are not listed beyond the international endpoint.