Every leading model, priced per result
21 models from the top labs, one credit balance. Pick the right one each time.
- OpenAI
- Black Forest Labs
- ByteDance
- Kuaishou
- Alibaba
- Hyper3D
- Tripo
- Microsoft
- Tencent
- Recraft
Image models
Generate and edit images. Priced per image.
| Model | Best at | From |
|---|---|---|
| Nano Banana 2.1GoogleGoogle's fast all-rounder for images with legible text and conversational edits. | Google's fast all-rounder for images with legible text and conversational edits. | 2per image |
| Nano Banana ProBestGoogleGoogle's most capable image model, for typography, infographics and precise edits. | Google's most capable image model, for typography, infographics and precise edits. | 6per image |
| GPT Image 2.5OpenAIOpenAI's image model, strongest at following detailed instructions and precise edits. | OpenAI's image model, strongest at following detailed instructions and precise edits. | 2per image |
| FLUX 3NewBlack Forest LabsBlack Forest Labs' newest model, for photoreal detail with up to 8 references. | Black Forest Labs' newest model, for photoreal detail with up to 8 references. | 1per image |
| Seedream 5 ProByteDanceByteDance's flagship image model, for crisp 2K images and reference-based edits. | ByteDance's flagship image model, for crisp 2K images and reference-based edits. | 2per image |
| Imagen 4 UltraGoogleGoogle's photorealistic text-to-image model for when fidelity matters most. | Google's photorealistic text-to-image model for when fidelity matters most. | 3per image |
| FLUX.2 KleinFastBlack Forest LabsSub-second drafts from Black Forest Labs, at 1 credit per image. | Sub-second drafts from Black Forest Labs, at 1 credit per image. | 1per image |
Video models
Text and image to video. Priced per clip, by length, resolution and audio.
| Model | Best at | From |
|---|---|---|
| Veo 3.1 FastGoogleVeo 3.1 quality with faster turnaround and a lower price, with audio. | Veo 3.1 quality with faster turnaround and a lower price, with audio. | 16per clip |
| Veo 3.1BestGoogleGoogle's flagship video model, with the highest fidelity and native audio. | Google's flagship video model, with the highest fidelity and native audio. | 32per clip |
| Kling 3.0KuaishouCinematic motion up to 15 seconds, with optional native audio. | Cinematic motion up to 15 seconds, with optional native audio. | 34per clip |
| Seedance 2.5NewByteDanceByteDance's newest video model, with synced dialogue, sound and long shots. | ByteDance's newest video model, with synced dialogue, sound and long shots. | 21per clip |
| Sora 2OpenAIOpenAI's video model, with synced audio in every clip. | OpenAI's video model, with synced audio in every clip. | 16per clip |
| Veo 3.1 LiteBudgetGoogleThe budget Veo, with native audio built in. Great for drafts and social clips. | The budget Veo, with native audio built in. Great for drafts and social clips. | 8per clip |
| Wan 3.0BudgetAlibabaLow-cost cinematic video up to 1080p, without audio. | Low-cost cinematic video up to 1080p, without audio. | 5per clip |
3D models
Image and text to textured GLB meshes. Priced per model.
| Model | Best at | From |
|---|---|---|
| Rodin Gen-2.5Hyper3DHyper3D's latest Rodin: crisp hard-surface geometry, quad meshes and PBR materials. | Hyper3D's latest Rodin: crisp hard-surface geometry, quad meshes and PBR materials. | 16per model |
| Rodin Gen-2.5 FastHyper3DRodin Gen-2.5's fast tier: light quad meshes with PBR, for a fraction of the price. | Rodin Gen-2.5's fast tier: light quad meshes with PBR, for a fraction of the price. | 4per model |
| Tripo H3.1TripoTripo's latest model: the closest match to your image, with sharp PBR textures. | Tripo's latest model: the closest match to your image, with sharp PBR textures. | 12per model |
| TRELLIS.2MicrosoftMicrosoft's open TRELLIS.2: faithful shapes with clean, remeshed topology. | Microsoft's open TRELLIS.2: faithful shapes with clean, remeshed topology. | 12per model |
| Rodin Gen-2Hyper3DHyper3D's production-ready quad meshes with PBR materials. | Hyper3D's production-ready quad meshes with PBR materials. | 16per model |
| Hunyuan3D 3.1TencentTencent's Pro image-to-3D model: precise, faithful geometry with PBR materials. | Tencent's Pro image-to-3D model: precise, faithful geometry with PBR materials. | 27per model |
Design models
Logos, icons and vectors.
| Model | Best at | From |
|---|---|---|
| Recraft V4.1RecraftRecraft's design model behind the logo and app icon tools, with true SVG output. | Recraft's design model behind the logo and app icon tools, with true SVG output. | 2per design |
Which model should I use?
Where we'd start for common jobs.
| If you need | Start with | Why |
|---|---|---|
| Text and layouts in the imagePosters, infographics and packaging with several lines of text. | Nano Banana Pro | Posters, infographics and packaging with several lines of text. |
| Photoreal people and productsNatural light and texture, up to 4K, with 8 references. | FLUX 3 | Natural light and texture, up to 4K, with 8 references. |
| Editing a photo you already haveDescribe the change; it keeps the rest of the picture. | Nano Banana 2.1 | Describe the change; it keeps the rest of the picture. |
| Long, detailed instructionsFollows many constraints in one prompt. | GPT Image 2.5 | Follows many constraints in one prompt. |
| Cheap drafts and exploration1 credit per image, a few seconds each. | FLUX.2 Klein | 1 credit per image, a few seconds each. |
| A hero video with soundHighest fidelity, native audio, start and end frames. | Veo 3.1 | Highest fidelity, native audio, start and end frames. |
| Social clips at volumeVeo quality with sound, faster and cheaper. | Veo 3.1 Fast | Veo quality with sound, faster and cheaper. |
| Shots longer than 10 secondsUp to 15 seconds with optional audio. | Kling 3.0 | Up to 15 seconds with optional audio. |
| Characters that talkDialogue in quotes, synced audio. | Seedance 2.5 | Dialogue in quotes, synced audio. |
| Silent b-roll on a budgetUp to 1080p for a low price per second. | Wan 3.0 | Up to 1080p for a low price per second. |
| A 3D asset for a game engineClean hard-surface geometry, quad meshes and PBR materials. | Rodin Gen-2.5 | Clean hard-surface geometry, quad meshes and PBR materials. |
| A vector logoTrue SVG output in the Logo Generator. | Recraft V4.1 | True SVG output in the Logo Generator. |
Questions
Something else? Ask us and a person will answer.
What is the best AI image generator?
It depends on the job. Nano Banana Pro is strongest with text and layouts, FLUX 3 with photorealism, GPT Image 2.5 with long instructions and Nano Banana 2.1 with fast edits. PhotoGenius lets you run any of them from the same studio, so you can compare on your own prompt.
What is the best AI video generator?
Veo 3.1 has the highest fidelity and native audio, Kling 3.0 handles longer shots up to 15 seconds, Seedance 2.5 is good with dialogue, Sora 2 generates sound in every clip and Wan 3.0 is the low-cost silent option.
How are models priced?
Each model has a fixed credit price per result that depends on its settings (resolution, length, audio). Plans start at $10 a month for 100 credits ($8 a month billed yearly), and the exact cost of every generation shows on the Generate button before you run it. Failed generations are refunded automatically.
Do I need separate subscriptions for Google, OpenAI and the others?
No. PhotoGenius runs every model through our providers, so one plan and one credit balance covers all of them.
How quickly are new models added?
We add major image, video and 3D models after testing them, and keep the older version only when it still has a reason to exist (for example a cheaper or faster option).