| Seedance 2.5 | Video model available through ByteDance products | Longer multimodal storytelling | Yes | 30s | Varies by access surface; not normalized here | Up to 30 images, 10 videos and 10 audio references; timestamp-level and reference-based editing | Not verified | Not verified | API access was still described as coming soon in the launch documentation | 2026-08-29 |
|---|
| Runway Gen-4.5 | Creative platform + video model | Professional creative workflows | No | 10s | 720p base output; multiple aspect ratios | Complex sequenced instructions, camera choreography and workspace iteration | Yes | 12 credits per generated second | Documented base generation output is 720p | 2026-08-29 |
|---|
| Kling AI | AI video platform + model family | Motion-focused creator workflows | Yes | 15s | Not normalized; varies by model and plan | Storyboard and multi-shot control documented for the VIDEO 3.0 series | Not verified | Not verified | Capabilities and access vary across the VIDEO 3.0 model series and plan | 2026-08-29 |
|---|
| Google Veo 3.1 | Video model available through Google products and APIs | Premium high-resolution audiovisual work | Yes | 8s | 720p, 1080p and 4K in supported Gemini API variants | First/last-frame and reference/extension features vary by model variant and surface | No | USD 0.4 per generated second | Capabilities, availability and resolution differ by Google surface and model variant | 2026-08-29 |
|---|
| Luma | Creative platform + video model | Fast production-oriented visual iteration | No | 10s | Draft, 540p, 720p and 1080p; HDR and EXR options | Performance tracking, keyframe/control and production pipeline features | Yes | USD 30 monthly subscription | Higher-fidelity and professional output modes consume substantially more credits | 2026-08-29 |
|---|
| Vidu Q3 | AI video platform + model family | Native audiovisual storytelling | Yes | 16s | 540p, 720p and 1080p on documented Q3 Pro/Turbo workflows | Smart scene cuts, frame-aware camera control and variant-dependent reference workflows | Yes | USD 0.045 per generated second | Workflow, reference and pricing details differ across Q3 Pro, Turbo, Mix and specialist variants | 2026-08-29 |
|---|
| MiniMax H3 outside Hailuo AI | Model + alternative access route | Keeping H3 while changing the product/access layer | Yes | 15s | Up to 2K in supported workflows | Multimodal text, image, video and audio context; workflow-dependent controls | Not verified | Not verified | Hosted, API and local workflows have different setup, pricing and hardware requirements | 2026-08-29 |
|---|