Models & Billing

What you'll be able to do after this module

Pick the right model for any job — image, video, audio, or LLM text. Estimate what your product will cost before writing a line of code. Avoid over-spending on quality you don't need.

How pricing works

Skytells uses flexible, unit-based pricing — each model is billed in the unit that best reflects how it consumes resources. There are no flat subscriptions or hidden compute surcharges; you pay only for what you use.

Billing units

Cost estimation examples

# Image model (unit: image)
500 images/day × $0.04/image × 30 days = $600/month

# Video model (unit: video_second)
10 clips/day × 8 seconds/clip × $0.12/second × 30 days = $288/month

# LLM text model (unit: million_token)
1,000,000 tokens/day × $2.00/million × 30 days = $60/month

Image models

Skytells' flagship image models are the TrueFusion family — purpose-built for quality, speed, and flexibility.

TrueFusion family

Third-party image models

Available on the same API with the same x-api-key header — no extra accounts needed:

Video models

Video predictions always take time (30s–5min). Use webhooks rather than polling. Output is always an MP4 URL.

Audio models

Text & LLM models

Text models power the Inference API (/v1/chat/completions, /v1/responses, /v1/embeddings). They are billed per token, not per prediction.

Choosing a text model

Fetching live model data

Always fetch pricing from the API rather than hardcoding it:

# List all models with pricing
curl https://api.skytells.ai/v1/models \
  -H "x-api-key: $SKYTELLS_API_KEY"

# Get a specific model
curl https://api.skytells.ai/v1/models/truefusion-pro \
  -H "x-api-key: $SKYTELLS_API_KEY"

Choosing the right model

Use this decision tree before picking a model:

Quick decision guide

Cost estimation worksheet

Fill this in before you build. Use the pricing object from GET /v1/models/:namespace to get unit and rate values programmatically:

def estimate_monthly_cost(
    units_per_day: float,
    cost_per_unit: float,
    days: int = 30,
) -> float:
    """Works for any billing unit — image, token, second, etc."""
    return units_per_day * cost_per_unit * days

# Image model (unit: image) — truefusion-pro at $0.04/image
print(estimate_monthly_cost(100,  0.04))      # $120/mo — 100 images/day
print(estimate_monthly_cost(500,  0.04))      # $600/mo — 500 images/day
print(estimate_monthly_cost(1000, 0.01))      # $300/mo — 1000 fast previews/day

# Video model (unit: video_second, 8 sec clips)
print(estimate_monthly_cost(10 * 8, 0.12))   # $288/mo — 10 clips/day × 8s × $0.12/s

# LLM text model (unit: million_token)
print(estimate_monthly_cost(1.0, 2.00))       # $60/mo — 1M tokens/day at $2/M

Rate limits by plan

When you hit a rate limit, the API returns 429 Too Many Requests with a Retry-After header. Implement exponential backoff in your retry logic.

Summary

You now know:

• The TrueFusion family and when to use each tier
• How to fetch live model data so your app stays accurate
• How to estimate monthly costs before you write a line of code
• truefusion-edge for real-time previews, truefusion-pro for final renders

Next steps:

• Image Generation path → — master prompts and integrate images in apps
• SDK Mastery → — use the Python or TypeScript SDK for cleaner code
• API Reference → — complete technical reference