The short version: Wan 3.0 is listed on Alibaba’s Qwen Cloud platform as wan3.0-video, with published pricing and an API example. The listing describes videos up to 30 seconds from text, image, audio and video inputs. The listing was checked on 9 October 2026 UTC (8 October in Vancouver); this update did not run a paid generation call.
The checked listing shows priced 480P, 720P and 1080P tiers, but no 4K tier. It does not establish whether Wan 3.0 weights have been released or how the model performs in independent tests. Those are different questions from what this page actually lists.
That gap between what Alibaba has published and what the internet says about it is the real story here, and it is worth more to you than another feature list.
What is verifiably true
Alibaba Cloud launched Qwen Cloud for international markets on 26 May 2026, announced in Singapore by Alibaba Cloud CTO and International President Dr. Li Feifei. It is an AI-native model marketplace aggregating more than 150 model APIs — Qwen, Wan and HappyHorse, alongside third-party models including GLM, Kimi and DeepSeek. Its Chinese counterpart operates as Qianwen AI.
This matters because Qwen Cloud is where Wan 3.0 appears. The checked first-party listing names wan3.0-video and shows an API example at https://maas.qwencloudapi.com/api/v1/services/aigc/video-generation/video-synthesis. That is a documented example, not evidence that Kingy successfully called it.
Here is what the listing showed at the October check. Prices are per second; the page displayed a 30% discount alongside its list prices. Eligibility, billing and the discount can change.
| Attribute | Value |
|---|---|
| Model ID | wan3.0-video |
| Inputs | Text, image, audio, video |
| Output | Video |
| Max duration | 30 seconds |
| Capabilities | Reference, editing, replication, driving |
| Pricing — 480P | $0.05 list / second; $0.035 displayed discounted / second |
| Pricing — 720P | $0.10 list / second; $0.07 displayed discounted / second |
| Pricing — 1080P | $0.20 list / second; $0.14 displayed discounted / second |
| Concurrency | 5 concurrent requests |
| Async queue | 500 tasks |
| Rate limit | 300 RPM |
Alibaba describes it as an all-in-one model with omni-modal reference, the ability to parse files, web pages and complex images, and “production-grade character consistency.”
The checked model page provides those specifications and its API example. It does not provide a technical report, architecture disclosure or parameter count. This check does not establish that such material, a model card or a formal announcement is absent elsewhere.
The checked listing has no 4K tier
The original August article examined search coverage claiming native 4K output at 3840×2160. Several pages led with cinematic 4K creation and comparison tables listing Wan 3.0 at “up to 4K UHD.” The examples below are retained from that historical check, not a new search-ranking audit.
The checked Qwen Cloud pricing table has three tiers: 480P, 720P and 1080P. No 4K price tier is shown on that page.
That gives buyers a concrete priced offer to inspect. It does not prove that 4K output is impossible or unavailable through every other route.
One of those sites is worth examining closely, because it is a useful specimen. Its hero banner reads “Coming Soon.” Its comparison table marks Wan 3.0 as shipped and current. Its body copy says Wan 3.0 is now available. Those three statements cannot all be true. Its use-case imagery for Wan 3.0 is served from /wan-2.7/ asset paths — recycled Wan 2.7 material, relabelled. Its comparison table claims Wan 2.5 and Wan 2.6 have no API access, which is flatly wrong; Alibaba has documented API access for both since 2025. And its own footer states it is not affiliated with, endorsed by or sponsored by any original model provider.
It is an affiliate wrapper optimising against pre-release search demand. Its meta keywords include “wan 3.0 release date” and “wan 3.0 waitlist” — it is built to capture people looking for a model before it launches.
What the checked listing says about weights
Weight availability needs a release, repository and licence check of its own. A commercial model listing is not a substitute for that evidence.
The Qwen Cloud model page checked for this update does not establish a Wan 3.0 weight release or licence. The earlier article described August repository searches; those searches are historical observations, not proof that no release exists elsewhere or now.
The original article examined pages asserting that Apache 2.0 weights shipped in April 2026 in 1.3B and 14B variants, traced that claim to a LinkedIn post, and reported that it did not find first-party corroboration. The 1.3B/14B pairing matched Wan 2.1’s February 2025 configuration; other pages claimed a 60B mixture-of-experts architecture. Those historical sourcing concerns do not establish what evidence or releases exist now.
The Wan open-weight record
The following release record is retained from the original August comparison. This update does not requalify its older weights or licence entries:
| Version | Released | Open weights |
|---|---|---|
| Wan 2.1 | Feb 2025 | Yes — Apache 2.0 |
| Wan 2.2 | Jul 2025 | Yes — Apache 2.0 |
| Wan 2.5 (preview) | Sep 2025 | No — API only |
| Wan 2.6 | Dec 2025 | No — API only |
| Wan 2.7 | Apr 2026 | No — API only |
| Wan 3.0 | 2026 listing; release date not established here | Not established by the checked model listing |
Wan 2.2 was the self-hosted option discussed in the original August comparison. This update does not establish that it remains Alibaba’s latest open-weight video release; check the current repository and licence before choosing a local deployment.
The nuance worth holding: Alibaba has not abandoned open weights generally. In July 2026 it released WanSong (text-to-music), Wan-Dancer-14B (music-to-dance) and Wan-Streamer (real-time streaming) under Apache 2.0. The strategy is legible — open the adjacent and ecosystem models, monetise the video flagship. Weigh any future “Wan will be open” pledge against that record.
What the historical benchmark comparison showed
The original August check did not identify Wan 3.0 among the 27 models inspected on Artificial Analysis Video Arena. That is a bounded, dated leaderboard observation, not proof that nobody has tested it or that it is absent from today’s rankings.
What we can see is where the rest of the field sits. Artificial Analysis text-to-video, with audio, as of early August 2026:
| # | Model | Creator | Elo | API $/min |
|---|---|---|---|---|
| 1 | Gemini Omni Flash | 1,244 | $6.00 | |
| 2 | MiniMax H3 | MiniMax | 1,238 | $7.80 |
| 3 | Dreamina Seedance 2.0 720p | ByteDance Seed | 1,224 | $9.07 |
| 4 | Wan2.7-260612 | Alibaba | 1,161 | $9.00 |
| 5 | HappyHorse-1.1 | Alibaba-ATH | 1,148 | $9.90 |
| 6 | HappyHorse-1.0 | Alibaba-ATH | 1,127 | $13.20 |
| 7 | Kling 3.0 1080p (Pro) | KlingAI | 1,111 | $20.16 |
| 8 | Wan 2.7 | Alibaba | 1,110 | $9.00 |
| 11 | Veo 3.1 | 1,093 | $24.00 |
Two things stand out.
The August table did not put Wan first. In that retained comparison, Alibaba’s best-scoring video model sat fourth, roughly 83 Elo behind the leader. Those results do not establish Wan 3.0’s quality or a current category winner.
Alibaba is hedging across two video lines. HappyHorse — an Alibaba-ATH model line — ranks immediately behind Wan 2.7 and ahead of it on some tasks. More tellingly, Alibaba’s own Model Studio documentation recommends HappyHorse over Wan as the default for image-to-video and reference-to-video work, steering users to Wan only when they need custom audio upload. When a vendor’s own docs route you away from its headline brand, that is worth noting.
The pricing problem
At the October check, Wan 3.0’s 1080P tier listed $0.20 per second and displayed a discounted $0.14 per second. Multiplying by 60 gives $12.00 per minute at list price or $8.40 per minute at the displayed discount. These are unit conversions, not a measured generation bill.
The original August table put Wan 2.7 at $9.00 per minute, MiniMax H3 at $7.80 and Gemini Omni Flash at $6.00. Those are retained historical observations. Comparing them with an October Wan price would mix dates, and the displayed Wan discount changes the earlier 33%-premium calculation.
The commercial question is whether Wan’s described 30-second generation fits your workflow at the price and conditions you can actually buy. A single-pass broadcast spot could have production value, but this update did not test continuity, quality or billing. It does not establish that Wan is more expensive than the current frontier.
The competitive set
The following competitive set preserves the original July/August 2026 comparison. Its launch claims, prices, rankings and API forecasts are historical; this October update checked Wan’s listing, not the current status of every competitor.
Seedance 2.5 (ByteDance)
Previewed at the Volcano Engine FORCE conference on 23 June 2026 and released 31 July 2026. ByteDance skipped versions 2.1 through 2.4 to signal a generational jump.
- 30-second single-pass audio-video generation, with multi-turn extension for longer sequences
- Up to 50 reference inputs — 30 images, 10 video clips, 10 audio clips — the highest input ceiling in the category
- Available on Jimeng (Dreamina) and Doubao Pro
- API not yet live. ByteDance’s own launch post says access is coming soon via BytePlus ModelArk; enterprise documentation points to 7 August 2026
- Closed weights
- No published pricing, resolution spec or benchmark claim in the primary announcement
Its predecessor, Seedance 2.0, sits third on the leaderboard and separately received a 4K upgrade at the same event.
MiniMax H3 (Hailuo 3.0)
Previewed at WAIC 2026 on 17 July and launched 31 July 2026 — the same day as Seedance 2.5.
- Native 2K (2560×1440), 24fps, 4–15 second clips, 32kHz stereo audio generated in the same pass
- Omni-reference: up to 9 images, 3 video clips, 3 audio clips — 12 files total
- #2 in text-to-video, #3 in image-to-video, #1 in video editing on Artificial Analysis
- $0.13 per second for 2K ($7.80/min); a 768p tier at $0.09/s is in closed beta
- Weights shipped 3 August 2026 — and this is the important part
MiniMax published H3 to Hugging Face as MiniMaxAI/MiniMax-H3: a 33B dense single-stream transformer, two task-specific checkpoints (FL2VA for frame-conditioned generation, Ref2VA for reference-driven), around 21GB each. The smallest working combination is 42.5GB, down 66% from 123.6GB at full precision. Native ComfyUI support merged the same day.
Three caveats that most coverage skipped:
- Local generation is natively 768px on the short edge, not 2K. The 2K output comes from a second in-context regeneration pass that remains hosted.
- H3-Context-IR and H3-Regenerate-2K did not ship. You get H3-Base; the full hosted system has three layers.
- The licence excludes the US, EU, UK and Korea from local deployment. The MiniMax H3 Community License, effective 2 August 2026, carves out those jurisdictions — citing the EU AI Act, UK and Korean rules, and ongoing US copyright litigation — alongside a $20M revenue threshold and an attribution requirement. MiniMax’s own repo documentation frames it as “not yet,” not “not ever.” The API remains globally available because MiniMax controls that infrastructure. Only the Qwen3-VL-32B text encoder ships under a genuine OSI licence (Apache 2.0).
If you are evaluating H3 for self-hosting, read the LICENSE file before you download anything. Canada is not on the exclusion list, but that is a determination to make against the actual document, not a blog post.
Head to head: October Wan listing and July/August competitor claims
| Wan 3.0 | Seedance 2.5 | MiniMax H3 | |
|---|---|---|---|
| Vendor | Alibaba | ByteDance | MiniMax |
| Released | Listing observed 8 October 2026 Pacific; release date not established | 31 Jul 2026 | 31 Jul 2026 |
| Max duration | 30s | 30s single-pass | 4–15s (extendable) |
| Resolution | Highest priced tier shown on checked listing: 1080P | Not published | 2K (2560×1440) |
| Native audio | Vendor-described audiovisual output; not generation-tested | Yes | Yes, 32kHz stereo |
| Reference inputs | Omni-modal, unspecified | 50 (30 img / 10 vid / 10 audio) | 12 (9 img / 3 vid / 3 audio) |
| API status | Qwen Cloud lists an API example; not called in this update | Pending (ModelArk) | Live |
| Price observations (different dates/resolutions) | October: $12.00 list / min; $8.40 displayed discount / min at 1080P | Not published | $7.80 (2K) |
| Open weights | Not established by checked listing | No | Yes, with carve-outs |
| Independent benchmark | No Wan 3.0 result identified in original August check; not refreshed here | None (2.0 ranks #3) | #2 T2V, #3 I2V, #1 editing |
The original comparison favoured MiniMax H3 on its then-reported ranking, price and downloadable checkpoints. That is a historical editorial judgment, not a current purchase recommendation. The October Wan listing supplies an API example and a 30-second description, but this update does not establish present competitor access, local licence eligibility or relative quality.
Wider context from the original August comparison
The original August Sora warning was time-sensitive. That section reported the consumer-app closure on 26 April 2026 and a scheduled Videos API shutdown on 24 September 2026, with no recommended replacement named in the cited notice. The deadline has passed; this historical warning is not current integration guidance, and this update did not recheck OpenAI’s notice.
Google’s best video model is not the one it markets as its video model. Gemini Omni Flash leads the arena at $6.00/min. Veo 3.1 — the branded flagship — ranks eleventh at $24.00/min: four times the price for lower measured quality.
Chinese labs hold most of the top ten. On the with-audio text-to-video leaderboard, eight of the top ten positions belong to Chinese labs. This is the structural fact underneath every “which video model” question in 2026, and most comparison content has not absorbed it.
How this was verified
Worth stating plainly, because the method is the point.
The original article attributed its comparison to Qwen Cloud, Alibaba Cloud Model Studio documentation, the Wan-AI and Wan-Video organisations, ByteDance’s launch materials, MiniMax’s model card and licence, and Artificial Analysis. The October update personally checked the exact Qwen Cloud Wan 3.0 listing for the prices, limits and documentation described above. It did not repeat every historical comparison or run generation tests.
Claims appearing only in aggregator content, affiliate sites or unattributed LinkedIn posts were excluded regardless of search ranking.
The checked listing does not substantiate the 4K specification, an April 2026 Apache 2.0 release, the 1.3B/14B parameter split or a 60B MoE architecture. That is a limit of this source check, rather than proof that no primary evidence for any of them can exist elsewhere.
The original August check documented this problem in the pages it reviewed: model-name domains carried specific pre-release claims that conflicted with the first-party record, and one inspected page reused Wan 2.7 asset paths while presenting Wan 3.0 claims. This update does not establish how often such pages now outrank primary documentation or whether every unsupported specification was invented. It establishes only that the checked claims lacked the first-party support described above.
Check the pricing page. It can identify what a vendor is offering and charging for at that moment. A missing price tier is a reason to ask for documentation, not proof that a capability does not exist.
What to watch
- Whether Alibaba publishes further Wan 3.0 documentation. The checked listing is commercial documentation, but this update does not establish a release date, technical report or current leaderboard position. Do not infer a staged rollout from those gaps alone.
- Whether a current Wan 3.0 benchmark can be inspected. A leaderboard result could add a useful quality signal, but the original August search cannot establish what independent evidence exists now.
- What happened after Seedance 2.5’s historical API forecast. The original comparison cited 7 August 2026. That date has passed; verify current documentation before treating the forecast as an access route.
- Whether a Wan 3.x weight release and licence can be verified. The original article discussed July ecosystem releases; the October model listing does not settle current downloadable-weight status.
- Whether MiniMax’s jurisdictional carve-outs become the norm. If open-weight video releases start shipping with geographic exclusions as standard, “open weights” stops meaning what practitioners assume it means.
Bottom line
Wan 3.0 has a first-party commercial listing describing a 30-second ceiling and broad input support. Buyers can inspect its listed tiers and limits. The listing alone does not establish generation quality, every available resolution or current weight availability.
If you want measured quality: inspect a current benchmark and its conditions.
If you want to self-host: verify the actual weights, hardware requirements and licence.
If you want 30-second generation: Wan’s listing describes it; test fit and billing before relying on it.
If you want the best answer: compare sources from the same date and distinguish a documented API example from a successful call.
Do not buy on a 4K claim that this checked listing does not substantiate. Ask for the exact resolution, endpoint and price before committing.
The Kingy Brief
Get the free AI price sheet.
25 AI plans and models on one page, from official pricing pages, updated every month. Free when you join The Kingy Brief: AI changes, original tests and one practical thing to try, Fridays at 09:00 Vancouver time.
Free · Double opt-in · Unsubscribe anytime
Signup help and newsletter schedule
After submitting, check your inbox for "Confirm your subscription to The Kingy Brief" and open its confirmation link. Your price sheet download is on the page that opens. Check Spam or Promotions if you cannot find it.
Fridays at 09:00 Vancouver time: source-checked AI changes, original tests and one practical thing to try. We skip a week when there is not enough verified material. Free. Unsubscribe anytime.