Wan 3.0 Prime Spicy vs Spicy: What the Price Gap Buys

Alibaba shipped six uncensored Wan 3.0 endpoints through Siray's API: alibaba/wan-3.0-{t2v,i2v,ref2v}-spicy and a Prime tier, alibaba/wan-3.0-{t2v,i2v,ref2v}-prime-spicy. Diff their OpenAPI request schemas and exactly one property differs: the model enum. Same fields, types, defaults, required list. The decision comes down to money and to output you judge yourself.

Key Takeaways

• Request bodies are field-for-field identical: model, prompt, duration, size, aspect_ratio, audio_enable, seed, prompt_expansion_enable. Neither tier has negative_prompt.
• Prime's list price is only about 12% above plain Spicy. What you actually pay differs by 33%, because the two tiers sit on different launch-month discounts (20% off vs 33% off).
• Because the schema is identical, switching tiers is a one-string change — which makes a controlled A/B cost $0.21.

The Request Bodies Are the Same, Field for Field

Both tiers require model, prompt, duration, size, and aspect_ratio, and both accept the same optional seed, audio_enable, and prompt_expansion_enable (default true). duration runs 2 to 30 whole seconds on each. size is 480p / 720p / 1080p on each. aspect_ratio offers the same six values on each: 16:9, 9:16, 1:1, 4:3, 3:4, adaptive. Neither tier exposes negative_prompt, so unwanted-element control is a positive-prompt problem on both.

Paying for Prime buys no capability: no longer clips, no extra resolution tier, no parameter the cheaper endpoint lacks. Whatever Prime is worth, it is worth in the pixels, not the API surface.

Where the 33% Actually Comes From

Prices below come from the console model-detail Pricing block on 2026-09-16, list price first, charged price second. Both tiers price linearly in duration: 720p is 2x the 480p rate, 1080p 4x.

Read the two columns of list prices, though, and the premium is roughly 12%, not 33%. The extra 21 points are a promotion artifact: Prime carries a 20% off badge, plain Spicy carries 33% off, both part of the Wan 3.0 launch-month campaign. Those badges have already moved twice this cycle. Re-read the Pricing block the day you run a long job rather than inheriting a number from an article, this one included.

Running the A/B for $0.21

Because the schema is identical, a controlled comparison is a single string swap. Pin seed, prompt, duration: 2, size: "480p", and aspect_ratio, then send the same body twice with different model values. That pair costs $0.12 + $0.09 = $0.21. Do it three times across the prompts your project actually uses and you have spent $0.63 answering the question for your own content, not for a tier name.

One control to set deliberately: prompt_expansion_enable defaults to true on both, so each tier rewrites your prompt before generating and you are no longer comparing the same instruction. Set it false for the A/B; decide about expansion in production separately.

Siray's normalized layer is what makes the swap a one-line change rather than two integrations, and the same key reaches plain Spicy, Prime Spicy, and the Seedance and MiniMax alternatives covered in Seedance 2.5 Spicy vs Wan 3.0 Spicy. For the plain tier's prompting behavior, see the Wan 3.0 Spicy uncensored video guide; for how the discounts have moved, Wan 3.0 launch week rates.

Developer Notes

Default to plain Spicy and treat Prime as an upgrade you have evidence for. Nothing in the spec argues the other way.

Ready to run that $0.21 comparison? Create your free Siray account and put both Wan 3.0 Spicy tiers against your own prompt.

A Note on Compliance

Siray's NSFW and uncensored models generate legal, consensual adult content only — CSAM and any illegal content are refused with zero tolerance. "Spicy" means no moderation beyond what the law requires, not that anything goes. Wan 3.0 Spicy and Wan 3.0 Prime Spicy are third-party Alibaba Cloud models available through Siray's API, not Siray-hosted models, and the same standard applies to everything generated with them.