Can Codex generate videos?
Seedance 2 team · Updated
Not by itself: Codex is OpenAI's coding agent and does not render video files. Add the Seedance 2 plugin or its MCP server (https://www.seedance2.net/mcp) and Codex generates Seedance 2.5, Kling 4.0, Veo 3.1 and other clips on your Seedance 2 account, uses images from your project as frames or references, and saves the results where you ask.
Steps
- Run codex plugin marketplace add nereo-io/seedance2-plugin, then codex plugin add seedance2@seedance2-marketplace, and start a new session.
- Sign in with your Seedance 2 account when Codex asks.
- Without the plugin: add [mcp_servers.seedance2] with url = "https://www.seedance2.net/mcp" to ~/.codex/config.toml.
Prompts to try
- Use ./public/hero.png as the first frame and make a 10-second Seedance 2.5 video at 1080p for the landing page. Save it to ./public/hero.mp4.
- For every image in ./products, make a 5-second Seedance 2.0 Fast clip with a slow push-in, 9:16, and save them to ./videos.
- Make a 2K Nano Banana 2.1 cover image for each post in ./content/blog that has none.
What happens when you ask
- The assistant calls list_models and tells you the credit price of the settings it picked.
- It calls generate_video, which returns a task id right away.
- It polls get_task until the video is done, usually one to several minutes, and gives you the file.
- If a generation fails, its credits are refunded automatically.
Video models Codex can use
Prices are in Seedance 2 credits, not part of your ChatGPT plan. Full tables: https://www.seedance2.net/pricing.md.
| Model | Modes | Length | Credits at default |
|---|---|---|---|
| Seedance 2 Mini | text-to-video, image-to-video, reference-to-video | 4–15 s | 3 |
| Seedance 2.0 Fast | text-to-video, image-to-video, reference-to-video | 4–15 s | 8 |
| Seedance 2.0 | text-to-video, image-to-video, reference-to-video | 4–15 s | 19 |
| Seedance 2.5 | text-to-video, image-to-video, reference-to-video | 4–30 s | 28 |
| Veo 3.1 Lite | text-to-video, image-to-video | 8–8 s | 6 |
| Veo 3.1 | text-to-video, image-to-video, reference-to-video | 8–8 s | 10 |
| Gemini Omni Flash 1.1 | text-to-video, image-to-video, reference-to-video | 4–10 s | 11 |
| MiniMax H3 | text-to-video, image-to-video, reference-to-video | 4–15 s | 12 |
| MiniMax H3 Max | text-to-video, image-to-video, reference-to-video | 5–15 s | 21 |
| MiniMax H3 Max Turbo | text-to-video, image-to-video | 5–15 s | 11 |
| Kling 4.0 | text-to-video, image-to-video, reference-to-video | 3–30 s | 35 |
| Kling 3.0 | text-to-video, image-to-video | 5–10 s | 25 |
| Wan 3.0 | text-to-video, image-to-video, reference-to-video | 2–30 s | 20 |
| Wan 3.0 Prime | text-to-video, image-to-video, reference-to-video | 2–30 s | 32 |
Questions
- Can Codex generate images?
- Yes, through the same Seedance 2 connection: Nano Banana 2.1 from 1 credit an image, GPT Image 2.6 and Seedream, for new images and for edits of files in your project.
- Does my ChatGPT plan pay for the videos?
- No. Generations are paid from your Seedance 2 credit balance. New accounts get free credits with no card, and failed generations are refunded automatically.
- Can Codex use my local images and audio?
- Yes. Codex uploads them with the upload_media tool and passes them to the model as first frames, last frames or references.