Veevid

Veevid — AI Video Generator

All-in-one AI video generator: text to video, image to video, and reference-to-video via Veevid AI. One API, 10+ state-of-the-art video generation models including Kling 3.0, Sora 2, Veo 3.1, LTX 2.3, Seedance 1.5 Pro, and Wan 2.6.

Setup

1. Get your API Key at veevid.ai/settings/api-keys
2. Save the key locally:

mkdir -p ~/.config/veevid
echo "YOUR_API_KEY" > ~/.config/veevid/api_key

What You Can Say

These phrases trigger this skill:

• "Generate a video of a sunset over the ocean"
• "Turn this image into a video"
• "Make a 10-second product ad with Kling 3.0"
• "Create a vertical video for TikTok"
• "How much does it cost to generate a video?"
• "What AI video models are available?"
• "Check my credit balance"

Image Retrieval from Chat (Discord / Group Chats)

When the user requests image-to-video but no image URL is provided in the message:

1. Use message read (channel=discord, limit=10) to scan recent messages in the current channel.
2. Filter attachments by the sender's user id to avoid picking up someone else's image.
3. Collect all recent image attachments (content_type starts with image/) from the sender, ordered newest first.
4. Preview by quoting: Use message send with replyTo set to the message id of the image message. This quotes the original message with its image preview inline, so the user can visually confirm. Add text like "Is this the right image?" Don't just show the filename — filenames like IMG_20260203_xyz.jpg are meaningless.
5. Single image needed (standard I2V): quote the most recent image message and ask the user to confirm.
6. Multiple images needed (Kling 3.0 start+end frames, Veo 3.1 reference-to-video with 1-3 refs): quote each image message with a number label and ask the user which ones to use and in what order.
7. If not enough images are found, ask the user to send the missing ones or provide URLs.

This ensures image-to-video works seamlessly even when Discord doesn't forward attachments directly to the agent. Applies to all models, not just a specific one.

Important: Always Confirm Before Generating

Before calling the generate API, ALWAYS:

1. Call POST /api/quote with the billing-relevant params for your planned generation (see Step 1 below) — it returns the precise credit cost and current balance.
2. Tell the user which model you'll use and the exact cost from the quote response.
3. For image-to-video: also confirm the image is correct (show filename/URL).
4. Wait for the user to confirm before proceeding.

If the generate API returns 402, report the exact required and balance values from the response body to the user.

If balance is insufficient, suggest a cheaper model or link to veevid.ai/pricing.

Step 1 — Quote (get exact cost before generating)

/api/quote only needs the billing-relevant params: mode, generation_type, duration, video_quality, aspect_ratio, and model-specific options (e.g. model_version, generate_audio). It does not require prompt, image, or other content fields — those don't affect credit cost.

curl -X POST https://veevid.ai/api/quote \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "mode": "veo3",
    "generation_type": "text-to-video",
    "video_quality": "standard"
  }'

Returns: {"required_credits": 20, "current_balance": 451, "sufficient": true}

If sufficient is false, suggest a cheaper model before asking the user to confirm.

Step 2 — Generate Video (Text to Video)

curl -X POST https://veevid.ai/api/generate-video \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "A golden retriever running through sunflowers, cinematic lighting",
    "mode": "veo3",
    "aspect_ratio": "16:9",
    "video_quality": "standard"
  }'

Returns either:

• Async case: {"generation_id": "vg_xxx", "status": "processing", "video_url": null, "image_url": null}
• Sync case: {"generation_id": "vg_xxx", "status": "completed", "video_url": "https://..."}

If status is already completed and a video_url is present, return it to the user immediately instead of polling.

Generate Video — Image to Video

If the user provides an online image URL, pass it directly:

curl -X POST https://veevid.ai/api/generate-video \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Gently animate this scene with camera movement",
    "mode": "veo3",
    "aspect_ratio": "16:9",
    "video_quality": "standard",
    "image": "https://example.com/photo.jpg",
    "generation_type": "image-to-video"
  }'

If the user provides a local file, the image URL isn't known yet — but credit cost doesn't depend on it, so quote first with billing params, then upload:

# Step 1: Quote (image URL not needed for pricing)
curl -X POST https://veevid.ai/api/quote \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{"mode": "veo3", "generation_type": "image-to-video", "video_quality": "standard"}'

# Step 2: Upload (only after user confirms)
curl -X POST https://veevid.ai/api/storage/upload \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -F "file=@/path/to/image.jpg"

# Returns: {"url": "https://cdn.veevid.ai/uploads/xxx.jpg", "key": "uploads/xxx.jpg"}

# Step 3: Generate with uploaded URL
curl -X POST https://veevid.ai/api/generate-video \
  -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  -H "Content-Type: application/json" \
  -d '{
    "prompt": "Animate this scene",
    "mode": "veo3",
    "video_quality": "standard",
    "image": "https://cdn.veevid.ai/uploads/xxx.jpg",
    "generation_type": "image-to-video"
  }'

Step 3 — Poll Status

If the generate response is still processing, poll until status is completed:

curl -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  https://veevid.ai/api/video-generation/{generation_id}/status

Status values: processing → completed or failed. Typical time: 60-180 seconds.

Check Credit Balance

curl -H "Authorization: Bearer $(cat ~/.config/veevid/api_key)" \
  https://veevid.ai/api/credits

Returns: {"success": true, "credits": 451, "balance": 451}

Available Models

Choose model based on the user's needs. Credit ranges depend on duration, resolution, and quality — use /api/quote for the exact cost before confirming with the user.

Quick Model Picker

Model Parameters

Each model has specific allowed values. See references/api-reference.md for the complete per-model parameter table.

Common parameters:

Error Codes

Example Conversations

Simple text-to-video:

User: "Make a 4-second video of a cat sleeping on a couch"
Agent: [calls /api/quote with mode=veo3, video_quality=standard]
Agent: "Note: Veo 3.1 always generates 8-second videos (duration is fixed). Cost: 20 credits. You have 311 credits. Proceed?"
User: "Yes"
Agent: [generates] "Done! Here's your 8-second video: [URL]."

Image-to-video:

User: "Turn this product photo into a short video" [attaches image]
Agent: [calls /api/quote with mode=veo3, generation_type=image-to-video]
Agent: "I'll upload your image and animate it with Veo 3.1. Cost: 20 credits. You have 200 credits. Proceed?"
User: "Go ahead"
Agent: [uploads image, generates] "Done! Here's your animated product video: [URL]"

Budget-conscious:

User: "Generate a video but keep it cheap"
Agent: [calls /api/quote with mode=grok-imagine, duration=6, video_quality=480p]
Agent: "The cheapest option is Grok Imagine at 10 credits for a 6-second 480p video. You have 50 credits. Want me to use that?"

Links

• Veevid AI
• Get API Key
• Pricing & Credits
• Blog & Tutorials