HunyuanVideo 1.5 - Upgrade Checklist for Production Teams

TL;DR
Treat HunyuanVideo 1.5 as a controlled upgrade, not a drop-in replacement.
Re-run a fixed benchmark pack, compare cost-per-usable-second, then phase rollout by use case.
Keep the original v1.0 baseline alive until quality and latency targets are met.

Context
This post is the version-specific companion to the original release overview:
HunyuanVideo — Tencent’s 13B-Parameter Open-Source AI Video (Research Overview).

1 Upgrade goal: measurable gains, not just newer weights

Before switching traffic to 1.5, define target deltas against your current v1.0 workflow:

• Prompt adherence (does output match intent without prompt overfitting?)
• Motion stability (fewer temporal artifacts across cuts and camera moves)
• Brand safety pass rate (fewer rejected generations)
• Throughput and cost (seconds per generation and usable output ratio)

If those deltas are not material in your environment, defer migration and keep v1.0 as default.

2 Four checks to run before production