Gemini Omni 1.2 Flash is a multimodal video generation model. It turns a text prompt, or a still image, into a short video clip of 3 to 15 seconds with native audio, in 720p or 1080p. / Gemini Omni 1.2 Flash 是多模態影片生成模型，可依文字提示或單張圖片生成 3 到 15 秒、含原生音訊的短片，提供 720p 與 1080p。
Playground
Performance
Last 24 hours, all hosts combined. If one host fails, BazaarLink automatically moves on to the next healthy one.
Hosts
The same model runs on several hosts, and each host's price and speed can differ. BazaarLink chooses by price, speed and stability, and switches to the next host when one misbehaves; you can also pin a host in the request.
| Host | Price per second | End-to-end latency | Availability (24h) |
|---|---|---|---|
 Google Vertex AIZDR | $0.07 | 1.44s | 100.0% |
 Google AI Studio | $0.07 | 5.14s | 100.0% |
Price comparison
| Resolution | With audio / s | Without audio / s | 4-second video |
|---|---|---|---|
| 720p | $0.07 NT$2.28 | $0.07 NT$2.28 | $0.278 NT$9.11 |
| 1080p | $0.104 NT$3.42 | $0.104 NT$3.42 | $0.417 NT$13.67 |
Video edits are billed as input seconds + output seconds; a failed job is refunded in full.
Quick start
Just change base_url to use Gemini Omni 1.2 Flash through the BazaarLink API.
# 1. Submit the job (returns vjob_xxx immediately)
curl https://api.bazaarlink.ai/v1/videos \
-H "Authorization: Bearer $BAZAARLINK_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "google/gemini-omni-1.2-flash",
"prompt": "a bird flying over mountains",
"duration": 4,
"resolution": "720p",
"generate_audio": false
}'
# → 202 { "id": "vjob_xxx", "status": "pending" }
# 2. Poll the status (at least 8 seconds apart)
curl -H "Authorization: Bearer $BAZAARLINK_API_KEY" \
https://api.bazaarlink.ai/v1/videos/vjob_xxx
# 3. Download the MP4 when it is done
curl -H "Authorization: Bearer $BAZAARLINK_API_KEY" \
-o output.mp4 \
https://api.bazaarlink.ai/v1/videos/vjob_xxx/contentFAQ
How long does a video take?
Videos are generated asynchronously: you get a vjob id right away and the video is usually done in 30 seconds to 5 minutes. Poll the status no more often than every 8 seconds.
How is it billed?
By resolution and seconds; the per-second prices are in the price table and differ with and without an audio track. The cost of a 4-second video is the last column of that table.
Am I charged if it fails?
No. Credit is reserved at submit time and refunded in full if the job fails or times out.
How do I download the video?
When the job is done, call /v1/videos/{id}/content with the same API key to download the MP4.
Explore
- Billed in USD with TWD quotes plus Taiwan e-invoices, so Taiwanese teams do not need an overseas card
- OpenAI-compatible API, no code rewrite
- When an upstream source has a problem, the request can try another available source; if none is available you get an error
- Chinese-language support from a local team