HappyHorse 1.0 Coming SoonLearn More
Wan API Pricing Guide: Wan 2.5, Wan 2.6, and Wan Image Costs on Evolink AI
guide

Wan API Pricing Guide: Wan 2.5, Wan 2.6, and Wan Image Costs on Evolink AI

EvoLink Team
EvoLink Team
Product Team
April 10, 2026
10 min read
If you are evaluating Wan 2.5, Wan 2.6, or Wan Image for a real workload, pricing is usually the first question your finance and product team will ask. This guide is a focused Wan API pricing guide: per-second video rates, per-image rates, the cheapest way to access Wan 2.5 today, and how Evolink AI compares to Alibaba's official DashScope pricing for the main Wan routes.
It is deliberately not a marketing page. If you want feature overviews and the playground, the Wan 2.6 model page, Wan 2.5 model page, and Wan Image model page cover that.
ModelModalityEvolink priceAlibaba official (reference)Where to verify
Wan 2.6 (standard tiers)Video, per second$0.0708 / sec$0.1019 / sec/wan-2-6
Wan 2.6 FlashVideo, per second$0.021-$0.069 / secVaries by variant/wan-2-6
Wan 2.5Video, per second$0.0708 / sec$0.1019 / sec/wan-2-5
Wan ImageImage, per image$0.0208 / image$0.0306 / image (reference)/wan-image
Cheapest Wan 2.5 access today: sign up for Evolink AI, use the available trial or introductory credits to validate quality on a few clips, then top up only after you confirm Wan 2.5 fits your workload. Treat the dashboard and API responses as the source of truth for live numbers - model pricing can change as Alibaba ships new variants.

1. Wan 2.6 API pricing

Wan 2.6 is Alibaba Tongyi Wanxiang's latest cinematic AI video generator, with three entry points: text-to-video (wan2.6-t2v), image-to-video (wan2.6-i2v), and reference-video (wan2.6-r2v). On Evolink AI, the standard Wan 2.6 text-to-video and image-to-video tiers are $0.0708 per second.
For Wan 2.6 reference video, budgeting is more nuanced: pricing depends on input duration + output duration, and 1080p carries a 1.67x quality multiplier. In practice, that means reference-video jobs should be estimated separately from standard text-to-video or image-to-video runs.
Compared with the repo's standard official reference rate of $0.1019 per second for the main Wan 2.6 route, Evolink AI's $0.0708 per second standard tier is roughly 30% cheaper. That comparison is most accurate for the standard Wan 2.6 tiers, not every Wan 2.6 variant.

Wan 2.6 Flash pricing

Wan 2.6 also has Flash variants for image-to-video and reference-video. Flash trades a small amount of quality for faster inference and a lower per-second cost, which makes it useful for:
  • A/B testing social ad hooks where you generate dozens of variants per concept
  • In-app video features where latency matters more than absolute quality
  • High-volume UGC pipelines that prioritize unit economics over cinematic polish
Based on the current repo pricing definitions, Wan 2.6 Flash runs in a range of roughly $0.021-$0.069 per second depending on settings. The standard rule of thumb: Flash is materially cheaper than standard Wan 2.6, but the standard model is the right choice when you want the full multi-shot cinematic output.

2. Wan 2.5 API pricing - and the cheapest access route

Wan 2.5 is the daily workhorse tier of the Wan video lineup. It is the model most teams reach for when they want short, story-like clips with audio output on current routes at a friendlier price point than the heavier cinematic models. On Evolink AI, the Wan 2.5 API price is $0.0708 per second, which puts it in the same per-second band as the main Wan 2.6 standard tier.
If you specifically searched "cheapest Wan 2.5" or "Wan 2.5 image to video free", the practical answer is:
  1. There is no fully free Wan 2.5 tier from either Alibaba or Evolink AI - Wan 2.5 is a paid AI video generator, not a free model.
  2. Trial or introductory credits are how you actually try it without committing. Evolink AI's current site messaging references free credits for new users, which is the practical way to test Wan 2.5 image-to-video or text-to-video before topping up.
  3. Compare per-second cost, not per-clip cost, when picking a route. A 5-second clip on Evolink AI at $0.0708/sec is approximately $0.354, but that figure changes with duration, resolution, and variant choice.
  4. Use the playground or preview first. The Wan 2.5 playground lets you validate output quality on your specific prompts before scaling up credit consumption.

This is the same advice we would give for any AI video API: validate first, then scale. Wan 2.5 specifically benefits from this approach because video quality varies more with prompt structure than with tier choice.


3. Wan Image API pricing

Wan Image is Alibaba Tongyi Wanxiang's image generation model, with both text-to-image and image-to-image (multi-image fusion) modes. It is billed per image, not per second.
On Evolink AI, Wan Image is $0.0208 per image, compared to a reference DashScope rate around $0.0306 per image. For batch creative workflows - campaign banners, catalog refreshes, A/B testing ad variants - this works out to significant savings as you scale into hundreds or thousands of generations.

Two practical notes:

  • Multi-image fusion costs the same as single-image generation in Wan Image's image-to-image mode, so you can combine product shots and backgrounds in one call without paying extra per reference.
  • Output count parameter lets you request multiple candidate images per call, and the cost scales linearly with that count. A common pattern is 3-4 candidates in the first pass, then narrower refinement batches.

Alibaba's official Wan API is exposed through DashScope and Model Studio, with regional console projects, China-region billing settings, and enterprise contracts. That works well if your stack is already on Alibaba Cloud, but it adds friction for teams outside mainland China.

Evolink AI provides a single unified Wan API endpoint for Wan 2.5, Wan 2.6, Wan 2.6 Flash, and Wan Image, with:
  • One clear standard per-second rate for Wan 2.5 and the main Wan 2.6 tiers, plus lower-priced Flash variants and separate reference-video budgeting logic
  • USD billing with credit-based top-ups (no enterprise contract required for small teams)
  • The same async task pattern across all four models, so your integration code stays consistent
  • Trial or introductory credits for new users - the practical "free trial" path for evaluating Wan before you commit budget
For most teams outside mainland China, the choice is straightforward: Evolink AI is usually cheaper on the standard Wan tiers, simpler to integrate, and easier to budget than calling DashScope directly. Teams already on Alibaba Cloud may prefer DashScope for billing consolidation reasons, which is a valid choice.

5. Is Wan open source? A note on Wan 2.5 and Wan 2.6

A common follow-up question on pricing pages is whether you can avoid API costs entirely by self-hosting Wan. The honest answer:

Alibaba open-sourced earlier Wan releases such as Wan 2.1, and you can find those weights on Hugging Face for self-hosted inference. Wan 2.5 and Wan 2.6 are documented as API-accessible models on Alibaba's DashScope and Model Studio. As of April 2026, we have not found an official Alibaba source confirming Wan 2.5 or Wan 2.6 themselves as open source - for the most current status, check Alibaba's official channels.

Practically: if you specifically need offline, self-hosted Wan and can accept the older Wan 2.1 quality, the open-source path is real. If you need the latest cinematic Wan 2.6 output or Wan 2.5 with audio output on current routes, the API route is the only available option today, and Evolink AI is one of the simplest mainstream ways to access it.


FAQ

Is Wan 2.6 API free?

No, Wan 2.6 is a paid AI video generator. You may be able to use trial or introductory credits on Evolink AI to generate your first Wan 2.6 clips before topping up, which is the closest thing to a free trial today. There is no permanent free tier.

What is the cheapest way to use Wan 2.5?

The cheapest practical route for most teams outside mainland China is to call Wan 2.5 through Evolink AI at $0.0708/sec, instead of going directly through Alibaba DashScope. Based on the current repo pricing references, that is roughly 30% lower than the standard official rate used for Wan 2.5 and the main Wan 2.6 tier.

How much does Wan Image cost per image?

Wan Image is $0.0208 per image on Evolink AI, compared to a DashScope reference rate around $0.0306 per image. Multi-image fusion in image-to-image mode does not cost extra per reference image.

Does Wan 2.6 Flash cost less than standard Wan 2.6?

Yes. Wan 2.6 Flash is meaningfully cheaper per second than the standard Wan 2.6 API, in exchange for slightly lower quality and faster inference. Based on the current repo pricing definitions, Flash sits around $0.021-$0.069/sec depending on settings. Check the Wan 2.6 dashboard for the live rate before budgeting production usage.

Can I get Wan 2.5 image-to-video for free?

There is no permanent free tier, but Evolink AI may offer trial or introductory credits that let you generate a few Wan 2.5 image-to-video clips before paying. After credits run out, billing is per-second at $0.0708/sec.

Is Wan open source so I can self-host?

Alibaba open-sourced earlier Wan releases such as Wan 2.1, but Wan 2.5 and Wan 2.6 are documented as API-accessible on Alibaba's DashScope and Model Studio. As of April 2026, we have not found an official Alibaba source confirming Wan 2.5 or Wan 2.6 themselves as open source. Check Alibaba's official channels for the most current status.

How is Wan 2.6 different from Wan 2.5 in terms of pricing?

On Evolink AI, both Wan 2.5 and the main Wan 2.6 standard tiers are priced at $0.0708 per second of generated video. The main differences are in the variant structure: Wan 2.6 also has Flash options at lower price points and reference-video modes with separate budgeting logic. In product terms, Wan 2.5 is the daily workhorse for short social clips, while Wan 2.6 is the cinematic multi-shot tier for longer narrative-style videos.


Get started

Sign up for Evolink AI to check current trial or introductory credit availability and test all three Wan models on real prompts before committing budget.

Ready to Reduce Your AI Costs by 89%?

Start using EvoLink today and experience the power of intelligent API routing.