Decision snapshot
| Decision factor | Wan 3.0 | Kling 3.0 1080p Pro |
|---|---|---|
| Text-to-video Elo | 1,240 | 1,107 |
| Per-minute cost at 1080p | $12.00 | $20.16 |
| Single-pass duration | 30 seconds | 15 seconds |
| Document input | Yes | No |
| Audio in single pass | Yes | No — separate TTS step |
| Open weights | No | Yes (predecessor line: Kling 2.5) |
Where each model wins
- Wan 3.0 wins on every axis except open weights — pick Wan 3.0 unless self-host is mandatory.
- Strict open-weight self-host — Kling 2.5 (predecessor) is the open-weight line.
- Document input — only Wan 3.0.
- 40 percent lower per-minute cost on 1080p — Wan 3.0.
Cost on the same workload
A 15-second 1080p clip costs $3.00 on Wan 3.0 versus $5.04 on Kling 3.0 1080p Pro. Wan 3.0 is 40 percent cheaper and 133 Elo ahead on the arena.
Ready to test the workflow?
Create account & add creditsWorkload recommendation
| Workload | Recommendation |
|---|---|
| Production 1080p video | Wan 3.0 — cheaper and higher ranked. |
| Self-host in your own VPC | Kling 2.5 (open weights, lower fidelity). |
| Document-to-video | Wan 3.0 — only it accepts documents. |
| 30-second single-pass | Wan 3.0 — Kling 3.0 is 15s. |
Benchmark rules
- Same harness version.
- Kling 3.0 1080p Pro is the tier in this table; lower tiers exist at lower price.
- Vendor-reported and independently measured results are separate classes.
- Do not extrapolate a single benchmark to production reliability.
Frequently asked questions
Can I try Wan 3.0 before integrating it?
Use the OneInfer Wan 3.0 launcher to open a prepared prompt in the public playground. Availability is checked against the current model catalog.
How should I treat benchmark claims?
Check the provenance label and harness version. Vendor-reported and independently verified results are deliberately shown as different evidence classes.
Put Wan 3.0 to work
Fund a controlled evaluation, send a reference frame or document, and measure quality and cost on your own workload.