Wan vs Veo
Wan and Veo are both current Top 10 AI video generators. Wan has the stronger current scoring signal in this dataset, while Veo may still win for specific access, workflow, pricing, or open-source needs.
Quick recommendation
Most ordinary users should start with Wan when they want the safer overall pick from the current evidence. Pick Veo instead when its winning scenarios below match your workflow better.
Choose Wan if
- price / costReason: Wan has the clearer cost/access advantage from open weights or confirmed price facts.Evidence: github.com, wan.video, gemini.google.com
- access / availabilityReason: Wan has broader visible access evidence across web, API, aggregator, or open-weight routes.Evidence: wan.video, gemini.google.com, labs.google
- open / closed sourceReason: Wan publishes open weights or open-source artifacts, which is a clear advantage for self-hosting, research, and reproducible evaluation.Evidence: github.com, deepmind.google
Choose Veo if
- camera controlReason: Veo has stronger camera control evidence from product features, workflow tags, or access data.Evidence: github.com, deepmind.google
- native audioReason: Veo has stronger native audio evidence from product features, workflow tags, or access data.Evidence: github.com, ai.google.dev
- API / developer availabilityReason: Veo has stronger API availability evidence from product features, workflow tags, or access data.Evidence: github.com, ai.google.dev
- version maturityReason: Veo has the newer tracked version event in the local version dataset.Evidence: artificialanalysis.ai, krea.ai, docs.dev.runwayml.com
Evidence pending before buying
Wan Overview
Alibaba's open and commercial video model family, with strong community adoption and variants for text, image, speech, and animation-driven video.
Company: Alibaba Group
Model page: Wan details
Fallback score: 87
Best Scenarios
Veo Overview
Google DeepMind's flagship video generation family for cinematic clips, image-referenced video, vertical output, native audio, and API workflows.
Company: Google DeepMind
Model page: Veo details
Fallback score: 86
Best Scenarios
Advantage Summary
Every dimension below names an advantage, a reason, evidence status, and the user type it matters for. Evidence pending means the page needs live pricing, leaderboard, benchmark, or product-term verification before naming a winner.
Where Wan wins
- price / costReason: Wan has the clearer cost/access advantage from open weights or confirmed price facts.Evidence: github.com, wan.video, gemini.google.comBest for: Budget-sensitive teams and API buyers
- access / availabilityReason: Wan has broader visible access evidence across web, API, aggregator, or open-weight routes.Evidence: wan.video, gemini.google.com, labs.googleBest for: Ordinary users who need the product to work today
- open / closed sourceReason: Wan publishes open weights or open-source artifacts, which is a clear advantage for self-hosting, research, and reproducible evaluation.Evidence: github.com, deepmind.googleBest for: Researchers, self-hosters, and teams needing model control
Where Veo wins
- camera controlReason: Veo has stronger camera control evidence from product features, workflow tags, or access data.Evidence: github.com, deepmind.googleBest for: Directors who need repeatable camera language
- native audioReason: Veo has stronger native audio evidence from product features, workflow tags, or access data.Evidence: github.com, ai.google.devBest for: Dialogue, sound effects, music, and audio-video sync
- API / developer availabilityReason: Veo has stronger API availability evidence from product features, workflow tags, or access data.Evidence: github.com, ai.google.devBest for: Developers, automation builders, and product teams
- version maturityReason: Veo has the newer tracked version event in the local version dataset.Evidence: artificialanalysis.ai, krea.ai, docs.dev.runwayml.comBest for: Teams choosing stable current products
Tie or depends
- feature depthReason: Feature coverage is close in the current profiles, so choose by the specific workflow rather than raw feature count.Evidence: github.com, arxiv.org, deepmind.googleBest for: Creators choosing by workflow breadth instead of one demo
- generation qualityReason: Fallback scores are close, so current use case and benchmark evidence matter more than the raw score.Evidence pending: needs live pricing, benchmark, or product documentation source.Best for: Teams optimizing final visual quality
No public source link is attached yet.
- speed / deliveryReason: Speed evidence pending; measure current queue time and generation latency before naming a winner.Evidence pending: needs live pricing, benchmark, or product documentation source.Best for: Users who need faster iteration or production throughput
No public source link is attached yet.
- motion realismReason: Benchmark evidence pending; use current leaderboard or controlled tests.Evidence: artificialanalysis.aiBest for: Ads, action shots, camera moves, and physical scenes
- prompt adherenceReason: Wan and Veo both have usable evidence here; choose based on workflow fit and current access.Evidence: github.com, deepmind.googleBest for: Prompt-heavy storyboards and precise scene requests
- subject consistencyReason: Both products have comparable subject consistency evidence or the difference depends on workflow depth.Evidence: github.com, deepmind.googleBest for: Character, product, and object continuity work
- reference controlReason: Both products have comparable reference control evidence or the difference depends on workflow depth.Evidence: github.com, deepmind.google, ai.google.devBest for: Brand/product references and character consistency
- editing / extensionReason: Both products have comparable editing or extension workflow evidence or the difference depends on workflow depth.Evidence: github.com, deepmind.googleBest for: Iterative editors and production teams
Advantage By Dimension
This is the decision layer: not just yes/no support, but who leads, why they lead, what evidence backs it, and who should care.
| Dimension | Advantage / Winner | Reason | Evidence | Best For |
|---|---|---|---|---|
| price / cost | Wan advantage | Reason: Wan has the clearer cost/access advantage from open weights or confirmed price facts. | Evidence: github.com, wan.video, gemini.google.com | Budget-sensitive teams and API buyers |
| access / availability | Wan advantage | Reason: Wan has broader visible access evidence across web, API, aggregator, or open-weight routes. | Evidence: wan.video, gemini.google.com, labs.google | Ordinary users who need the product to work today |
| feature depth | Tie or depends | Reason: Feature coverage is close in the current profiles, so choose by the specific workflow rather than raw feature count. | Evidence: github.com, arxiv.org, deepmind.google | Creators choosing by workflow breadth instead of one demo |
| generation quality | Tie or depends | Reason: Fallback scores are close, so current use case and benchmark evidence matter more than the raw score. | Evidence pending: needs live pricing, benchmark, or product documentation source. No public source link is attached yet. |
Teams optimizing final visual quality |
| speed / delivery | Evidence pending | Reason: Speed evidence pending; measure current queue time and generation latency before naming a winner. | Evidence pending: needs live pricing, benchmark, or product documentation source. No public source link is attached yet. |
Users who need faster iteration or production throughput |
| motion realism | Evidence pending | Reason: Benchmark evidence pending; use current leaderboard or controlled tests. | Evidence: artificialanalysis.ai | Ads, action shots, camera moves, and physical scenes |
| prompt adherence | Tie or depends | Reason: Wan and Veo both have usable evidence here; choose based on workflow fit and current access. | Evidence: github.com, deepmind.google | Prompt-heavy storyboards and precise scene requests |
| subject consistency | Tie or depends | Reason: Both products have comparable subject consistency evidence or the difference depends on workflow depth. | Evidence: github.com, deepmind.google | Character, product, and object continuity work |
| camera control | Veo advantage | Reason: Veo has stronger camera control evidence from product features, workflow tags, or access data. | Evidence: github.com, deepmind.google | Directors who need repeatable camera language |
| reference control | Tie or depends | Reason: Both products have comparable reference control evidence or the difference depends on workflow depth. | Evidence: github.com, deepmind.google, ai.google.dev | Brand/product references and character consistency |
| native audio | Veo advantage | Reason: Veo has stronger native audio evidence from product features, workflow tags, or access data. | Evidence: github.com, ai.google.dev | Dialogue, sound effects, music, and audio-video sync |
| editing / extension | Tie or depends | Reason: Both products have comparable editing or extension workflow evidence or the difference depends on workflow depth. | Evidence: github.com, deepmind.google | Iterative editors and production teams |
| API / developer availability | Veo advantage | Reason: Veo has stronger API availability evidence from product features, workflow tags, or access data. | Evidence: github.com, ai.google.dev | Developers, automation builders, and product teams |
| commercial use | Evidence pending | Reason: Commercial use depends on current product terms, regional plans, and license language; verify live terms before recommending a winner. | Evidence: github.com, deepmind.google | Brands, agencies, and revenue-generating projects |
| open / closed source | Wan advantage | Reason: Wan publishes open weights or open-source artifacts, which is a clear advantage for self-hosting, research, and reproducible evaluation. | Evidence: github.com, deepmind.google | Researchers, self-hosters, and teams needing model control |
| version maturity | Veo advantage | Reason: Veo has the newer tracked version event in the local version dataset. | Evidence: artificialanalysis.ai, krea.ai, docs.dev.runwayml.com | Teams choosing stable current products |
| external ratings / leaderboard | Tie or depends | Reason: Fallback scores are close, so current use case and benchmark evidence matter more than the raw score. | Evidence pending: needs live pricing, benchmark, or product documentation source. No public source link is attached yet. |
Buyers who want market validation |
| SEO / popularity / web authority | Evidence pending | Reason: SEO, popularity, and authority metrics need richer imports before naming a winner. | Evidence pending: needs live pricing, benchmark, or product documentation source. No public source link is attached yet. |
Teams using adoption and discoverability as risk signals |
| open-source GitHub metrics | Evidence pending | Reason: GitHub metrics only apply when a model has open-source or open-weight repositories; neither side has enough repo evidence here. | Evidence pending: needs live pricing, benchmark, or product documentation source. No public source link is attached yet. |
Open-model evaluators and infra teams |
| company background | Tie or depends | Reason: Both companies have usable profile evidence; brand strength alone should not decide the product choice. | Evidence: home.alibabagroup.com, deepmind.google, github.com | Users who care about product longevity, distribution, and support risk |
Feature Dimension Comparison
Supported values are source-backed where product detail records include source URLs; inferred values are marked.
| Dimension | Wan | Veo |
|---|---|---|
| text-to-video | Supported | Supported |
| image-to-video | Supported | Supported |
| video-to-video | Partial Animate/replacement workflows use video inputs; general V2V coverage depends on wrapper. | Supported Veo 3.1 supports video input for extension in Gemini API docs. |
| reference control | Supported | Supported |
| native audio | Partial Speech-to-video supports audio-driven generation; general native generated audio for T2V needs separate confirmation. | Supported |
| lip sync | Supported S2V audio-driven generation covers speech/video synchronization workflows. | Partial Native audio is confirmed; standalone lip-sync controls need feature-level confirmation. |
| camera control | Partial | Supported |
| editing/extension | Partial | Partial |
| API | Partial Self-hosting and hosted wrappers exist; Alibaba Cloud commercial endpoints should be normalized separately. | Supported |
| commercial license | Provisional Check Apache-2.0 license before commercial use. No public source link is attached yet. | Provisional Commercial use depends on current hosted product terms. No public source link is attached yet. |
| open/closed source | [Object Object] Wan2.2 publishes code and model weights with Apache-2.0 repository licensing shown by GitHub repository metadata. | [Object Object] Veo weights are not open. |
Wan Pricing / Access
Open weights can be downloaded, but practical cost depends on GPU hardware or hosted inference. Commercial Alibaba/partner endpoints should be priced separately.
If exact public pricing is not listed here, treat it as provisional and check the linked source before publishing a cost claim.
Access Channels
Price Facts
- Self-hosting costHardware-dependent; S2V examples note high VRAM requirements.
Veo Pricing / Access
Gemini API pricing lists Veo 3 at USD 0.40 per second, Veo 3 Fast at USD 0.15 per second, and Veo 2 at USD 0.35 per second; check page for Veo 3.1 pricing before display.
If exact public pricing is not listed here, treat it as provisional and check the linked source before publishing a cost claim.
Access Channels
Price Facts
- Veo 3 APIUSD 0.40 per second on paid tier
- Veo 3 Fast APIUSD 0.15 per second on paid tier
Performance / Quality Evidence
Rows say source-backed when comparison JSON provides a cited claim; otherwise they are provisional.
| Quality Dimension | Wan | Veo |
|---|---|---|
| motion realism | source-backed high Wan has a current evidence-weighted score of 87.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) | source-backed high Veo has a current evidence-weighted score of 86.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) |
| prompt adherence | source-backed high Wan has a current evidence-weighted score of 87.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) | source-backed high Veo has a current evidence-weighted score of 86.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) |
| temporal consistency | source-backed high Wan has a current evidence-weighted score of 87.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) | source-backed high Veo has a current evidence-weighted score of 86.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) |
| subject consistency | source-backed high Wan has a current evidence-weighted score of 87.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) | source-backed high Veo has a current evidence-weighted score of 86.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) |
| speed | source-backed high Wan has a current evidence-weighted score of 87.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) | source-backed high Veo has a current evidence-weighted score of 86.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) |
| resolution/duration/public benchmark evidence | source-backed high Wan has a current evidence-weighted score of 87.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) | source-backed high Veo has a current evidence-weighted score of 86.4. Treat this as source-aware provisional unless refreshed against live Artificial Analysis, Arena, and product tests. (high) |
Version Progress
Latest and historical versions are pulled from product profiles plus version-events JSON.
Wan versions
- Wan 2.5 / 2.6 / 2.7Names appear in third-party catalogues and should not be treated as confirmed official releases yet.
- Wan2.2Released inference code and model weights. Supports T2V, I2V, and TI2V. Integrated with Diffusers and ComfyUI.
- Wan technical reportPublished the Wan open large-scale video generative model paper line.
Veo versions
- Veo negative prompt supportRunway API adds optional negativePrompt for veo3, veo3.1, and veo3.1_fast text-to-video and image-to-video requests.
- Veo model cardsOfficial DeepMind model-card index for Veo safety/version updates.
- Veo 3.1 Lite model cardGoogle DeepMind published the Veo 3.1 Lite model card for the text/image-to-video system with audio.
- Veo 3.1 in Google Vids free monthly accessGoogle announced that personal Google accounts get monthly no-cost Veo 3.1 video generations in Google Vids, with higher limits for Google AI Pro/Ultra and Workspace AI Ultra accounts.
- Veo 3.1Improves Ingredients to Video. Adds native vertical outputs. Adds 1080p and 4K upscaling options.
- Veo 3.1 Preview / Fast PreviewGemini API model codes include veo-3.1-generate-preview and veo-3.1-fast-generate-preview. Supports text, image, last-frame, reference-image, and video-extension inputs depending on model.
- Veo 3 PreviewGemini API changelog lists video with audio generation for Veo 3 preview.
- VeoVeo announced at Google I/O 2024 as Google DeepMind's generative video model.
External Ratings
Fallback scores, SEO/web authority signals, GitHub status, and preset editorial winners are shown with caveats.
| Signal | Wan | Veo |
|---|---|---|
| Fallback score | 87 Wan appears in Artificial Analysis and Arena snapshots, and the open repository/paper provide reproducible public evidence.; professional score is provisional/evidence pending unless backed by scoring JSON. | 86 Google Gemini Omni Flash/Veo-family entries lead the Artificial Analysis audio-video snapshot and rank highly in Arena/Vercel snapshots.; professional score is provisional/evidence pending unless backed by scoring JSON. |
| SEO / web authority | Reachability collected; third-party authority metrics pending. | Reachability collected; third-party authority metrics pending. |
| GitHub / open source | Wan2.2 publishes code and model weights with Apache-2.0 repository licensing shown by GitHub repository metadata. | Veo weights are not open. |
| Comparison preset | Open Weight Benchmarking; Developer Workflows; Image To Video; Open Model Ecosystem; MoE Video Diffusion Newer Wan 2.5/2.6/2.7 names need official release-page confirmation.; Open checkpoints can require high-memory GPUs. | Cinematic Generation; Image Referenced Clips; Vertical Short Form Video; Ingredients To Video; Native Vertical Output Model availability and feature access vary by country, product surface, and plan.; Some Google products may expose only selected Veo versions. |
Which should you choose?
- Wan best-fit workflows: Wanopen-weight benchmarking, developer workflows, image-to-video
- Veo best-fit workflows: Veocinematic generation, image-referenced clips, vertical short-form video
- Score-backed shortlist pick: WanScores are close (87.4 vs 86.4), so run a prompt-level test for the exact use case.
Sources
Official sources, news/version sources, external ranking or authority sources, and open-source repository links used by this generated page.