Cost model docs
Cost visibility is an API feature, not a billing afterthought.
Ayneye exposes cost as an artifact so developers can design product limits before usage surprises them. The cost model should be visible in docs, dashboard, and API responses.
Cost artifact fields
| Field | Role | Meaning |
|---|---|---|
| video_hours_processed | Core unit | Bounded duration processed into artifacts |
| capture_seconds | Execution cap | Prevents accidental unbounded stream processing |
| detector_mode | Compute signal | CPU mode, fallback mode, or review-required state |
| retention_days | Storage signal | How long artifacts stay available |
| ask_queries | Downstream usage | How often the artifacts are queried |
| review_flags | Trust signal | Where human review is required |
How teams should evaluate cost
Run one easy clip, one hard clip, and one representative customer clip. Inspect cost.json after each execution. Then ask multiple questions against the same artifacts and compare the repeated-query path with workflows that repeatedly send visual context to a model. The answer should not be just a cheaper number; it should include evidence quality, confidence, latency, retention, and review load.
curl https://api.ayneye.com/api/product/videos/$VIDEO_ID/cost \
-H "Authorization: Bearer $AYNEYE_API_KEY"Cost policy
Keep limits visible, return clear limit errors, expose fallback and review-required states, and never hide cost drivers behind vague AI usage language.