AI video and image models in Dora — comparison and pricing
Dora gives you access to 13+ AI models from ByteDance, Google, OpenAI, xAI, Kuaishou and ElevenLabs from one subscription. Below is a comparison of every model — type, provider, output spec, and cost in credits.
Comparison table
| Model | Provider | Type | Output | Cost (credits) |
|---|---|---|---|---|
| SeeDance 2 | ByteDance | Video | 720p / 8s | 328 |
| SeeDance 2 Fast | ByteDance | Video | 480p / 8s | 124 |
| Seedance 2 Mini | ByteDance | Video | up to 720p / 12s | 16 (480p, 4s; 31 for 8s, 66 for 8s at 720p) |
| Gemini Omni | Video | 720p–4K / 4–10s · any-to-any | 120 (720p/1080p, 6s; 4K from 210; video-input mode is flat 240/360) | |
| Kling 2.6 | Kuaishou | Video | image-to-video, optional audio | 55 (5s without audio; 110 credits with audio) |
| Grok | xAI | Video | 480p / 6s | 10 |
| Nano Banana Pro | Image | up to 4K | 18 (1K/2K; 24 credits at 4K) | |
| Nano Banana 2 | Image | up to 4K | 10 | |
| Nano Banana | Image | standard | 4 | |
| GPT-4o Image | OpenAI | Image | standard | 6 |
| ChatGPT 2 | OpenAI | Image | standard | 6 |
| Dora | Dora Studio | Image | standard | 1 |
| ElevenLabs SFX v2 | ElevenLabs | Audio | AI sound effects from text | 5 |
SeeDance 2
- Provider: ByteDance
- Type: Video
- Output: 720p / 8s
- Cost: 328 credits
- Best for: Highest-quality cinematic video.
When to use
SeeDance 2 is the highest-quality video model in Dora as of April 2026. It produces an 8-second 720p clip in roughly 60 seconds with strong character consistency across reference frames and cinematic motion. Use it when output quality matters more than turnaround time or cost.
Limitations
Slow and expensive: 328 credits per 8-second clip. No 1080p output. Reference-frame consistency degrades after the third frame.
SeeDance 2 Fast
- Provider: ByteDance
- Type: Video
- Output: 480p / 8s
- Cost: 124 credits
- Best for: Half the cost of SeeDance 2, similar look.
When to use
SeeDance 2 Fast trades resolution for speed and cost. The 480p output is suitable for social previews, story-mode reels, and iteration. Quality and motion are nearly identical to SeeDance 2 — just at half the resolution.
Limitations
480p only; not suitable for delivery on widescreen or high-DPI displays. 124 credits per 8-second clip.
Seedance 2 Mini
- Provider: ByteDance
- Type: Video
- Output: up to 720p / 12s
- Cost: 16 credits (480p, 4s; 31 for 8s, 66 for 8s at 720p)
- Best for: The cheapest SeeDance, with sound included. Dora's default video model.
When to use
Seedance 2 Mini replaced SeeDance 1.5 Pro, which its provider retired on October 25, 2026; requests for 1.5 Pro in Dora now go to Mini. In a same-prompt test of five budget models it kept the selfie framing and the face from the reference photo, at about the same price per second as Veo 3.1 Lite (see the test on the blog: /blog/cheap-ai-video-models-same-prompt-test).
Limitations
No 1080p, and one reference photo (no last frame). 480p looks soft on large screens; use 720p for the final version.
Gemini Omni
- Provider: Google
- Type: Video
- Output: 720p–4K / 4–10s · any-to-any
- Cost: 120 credits (720p/1080p, 6s; 4K from 210; video-input mode is flat 240/360)
- Best for: Any-to-any video: from text, reference photos, or a video clip.
When to use
Gemini Omni (Gemini Omni Flash) is Google's any-to-any model: it generates video from a text prompt, up to 7 reference images, or transforms an existing video clip — all from one model. That flexibility makes it the best default when your inputs vary. It supports 720p, 1080p and 4K output, durations of 4–10 seconds, and 16:9 or 9:16. The video-input mode is especially useful for restyling or transforming footage you already have, rather than only generating from scratch.
Limitations
Only 16:9 and 9:16 aspect ratios. In video-input mode the output duration is auto and uses the first 10 seconds of the source clip. 4K costs more (210–300 credits) than 720p/1080p (90–180).
Kling 2.6
- Provider: Kuaishou
- Type: Video
- Output: image-to-video, optional audio
- Cost: 55 credits (5s without audio; 110 credits with audio)
- Best for: Image-to-video with native audio generation.
When to use
Kling 2.6 is image-to-video only — you provide a single source frame and it generates motion. The optional audio flag adds a generated soundtrack matched to the video. Best for animating a still image with sync sound in one pass.
Limitations
Cannot start from text; requires a reference image. Audio adds 100% cost: 110 credits for 5s with audio.
Grok
- Provider: xAI
- Type: Video
- Output: 480p / 6s
- Cost: 10 credits
- Best for: Cheapest text-to-video; Spicy mode for adult-adjacent content.
When to use
Grok by xAI is the fastest and cheapest video option. Three modes — fun, normal, spicy — let you tune the output style. The cheapest video generation in Dora at 10 credits.
Limitations
480p / 6s ceiling; output quality below SeeDance 2 Fast. Spicy mode output guidelines vary; check your platform of distribution before publishing.
Nano Banana Pro
- Provider: Google
- Type: Image
- Output: up to 4K
- Cost: 18 credits (1K/2K; 24 credits at 4K)
- Best for: Hyper-realistic images at 4K.
When to use
Nano Banana Pro is the highest-quality image model in Dora — 4K output, photorealistic faces and textures. Best when the image is a deliverable rather than a thumbnail or preview. 18 credits for 1K/2K, 24 credits at 4K.
Limitations
4K is roughly 30 seconds of compute; 2K is the practical sweet spot. 24 credits at 4K.
Nano Banana 2
- Provider: Google
- Type: Image
- Output: up to 4K
- Cost: 10 credits
- Best for: Best-quality image generation at lower cost than Nano Banana Pro.
When to use
Nano Banana 2 sits between standard Nano Banana and Nano Banana Pro. 10 credits, fewer aspect-ratio limitations than Pro, suitable for most editorial and product imagery.
Limitations
Slightly lower facial fidelity than Nano Banana Pro at 4K.
Nano Banana
- Provider: Google
- Type: Image
- Output: standard
- Cost: 4 credits
- Best for: Cheap, fast images for ideation.
When to use
Nano Banana is the entry-level image model — 4 credits, fast turnaround. Use for sketching, mood-boarding, prompt iteration before committing to a Pro/Pro 2 generation.
Limitations
Standard resolution only. Less detail in faces and textures than the Pro variants.
GPT-4o Image
- Provider: OpenAI
- Type: Image
- Output: standard
- Cost: 6 credits
- Best for: Smart text rendering and on-image typography.
When to use
GPT-4o Image excels at rendering legible text within the image — signs, posters, UI mockups. Most other image models produce illegible text. Use this when the image needs to contain readable words.
Limitations
Lower photorealism than Nano Banana Pro for portraits.
ChatGPT 2
- Provider: OpenAI
- Type: Image
- Output: standard
- Cost: 6 credits
- Best for: OpenAI's newer image generation model.
When to use
ChatGPT 2 (GPT-Image-2) is the newer OpenAI image generator. Similar pricing to GPT-4o Image but improved adherence to complex prompts and better handling of multiple subjects.
Limitations
Same standard resolution as GPT-4o Image. 6 credits.
Dora
- Provider: Dora Studio
- Type: Image
- Output: standard
- Cost: 1 credits
- Best for: Fast in-house AI images at 1 credit each.
When to use
Dora is the in-house image model: 1 credit per image on every plan. Subject to a fair use policy to prevent abuse.
Limitations
Standard resolution only. Best for thumbnails, social previews, prompt exploration.
ElevenLabs SFX v2
- Provider: ElevenLabs
- Type: Audio
- Output: AI sound effects from text
- Cost: 5 credits
- Best for: Royalty-free sound effects from a text prompt.
When to use
ElevenLabs Sound Effects v2 generates short audio clips ("explosion with debris", "footsteps on wood") from a text description. Output is 48 kHz, royalty-free, and the model decides duration. 5 credits per generation.
Limitations
Auto-duration only in Dora — you cannot specify a target length. Best for short SFX, not music.
FAQ
Which Dora model is fastest?
Grok and Dora (in-house) are the fastest options — Grok generates a 6-second 480p video in roughly 30 seconds; Dora generates an image in under 5 seconds. For higher-quality video, SeeDance 2 Fast finishes in roughly 30-45 seconds.
Which AI video models support audio generation?
Seedance 2 Mini makes sound by default at no extra cost (16 credits for 4 seconds at 480p). Kling 3.0 Omni generates native sound (90 credits for a 5-second clip at 720p). For standalone sound effects, ElevenLabs SFX v2 generates 48 kHz audio from a text prompt at 5 credits per generation.
What's the cheapest way to generate AI video in Dora?
Grok at 10 credits is the cheapest text-to-video. For animating a still image, Seedance 2 Mini at 480p / 4s costs 16 credits, sound included.