Soutine AI

Soutine AI is an AI image and video generator that brings multiple leading models together into one creative workspace. Users can create based on prompts or reference images, and every generation result is saved in a shared history for easy download, refinement, upscaling, or continuing into related workflows.

【Main Capabilities】 According to the official page, Soutine AI provides the following capabilities around image and video generation:

• Image generation: create images via prompts, or edit or transform based on reference images.
• Video generation: describe scenes, actions, and camera movements with prompts to generate videos; you can also extend an existing image into a continuous shot.
• AI video upscaling: upscale clips to 4K and restore sharp details.
• Image enhancement and restoration: the page mentions "Enhance & Restore".
• Reference-image editing: upload a reference photo and describe the desired modifications; some models can preserve the person's identity, pose, and scene details while editing only the specified parts. The official site showcases before-and-after examples with GPT Image 2.5, such as changing a red sweater into a white tuxedo while keeping the face, pose, printed photo, and room environment unchanged; turning a pet photo into an action-movie stunt double; and tidying up a room and making the bed in the same bedroom.
• Prompt and style library: the page offers up to 1000+ Nano Banana prompts and styles, covering portraits, anime, photo editing, and commercial design, with one-click copy for use.

【Available Models】 The officially listed models include Kling 4, Qwen Image 2.1 (New), Nano Banana, Nano Banana 2 Lite, Nano Banana Pro, GPT Image 2.5 (New). The page has a "View All Models" entry. The models and settings available under different tasks change accordingly; Soutine AI displays the models compatible with the current task and updates image and setting limits accordingly.

【How to Use】 The official process is three steps:

1. Describe or upload: describe what you want to create in Soutine AI, or add a reference image when the selected model supports it.
2. Choose a model for the task: select from the compatible models available for the current workflow; Soutine AI will update the settings and image limits according to the selected model. Standard reference-image input supports PNG, JPEG, and WebP.
3. Generate and continue processing: tasks are saved in the shared history and can be downloaded, refined, upscaled, or continued into related workflows. Generation time depends on the model, media type, resolution, duration, and queue load. After submission, tasks can be tracked, and results will appear in the corresponding generator and "My Creations." The credit cost is shown before generation and may vary depending on the model, task, resolution, duration, and input media. Eligible system or provider failures automatically refund the deducted credits. If a task remains in a pending state for a long time, it is recommended to avoid resubmitting it and to contact support with the task details.

[Use Cases] The scenarios shown on the page include:

• Converting illustrations into character figurine scenes, for example, placing a box printed with a character image in an indoor scene, a computer screen showing the Blender modeling process, and a character figurine standing on a circular plastic base.
• Making local modifications to a reference photo while preserving the person's identity, pose, and scene details.
• Prompt-driven visual creation such as portraits, anime, photo editing, and commercial design.
• Extending a still image into a continuous video shot.
• Upscaling video clips to 4K and restoring details.

[Billing and Membership] The official page shows that you can start for free, then upgrade as needed, and cancel at any time.

• Starter: $214.80 per year, automatically renews, cancel at any time. Automatic image generation 40% OFF; other base models 40% OFF; SOTA models 20% OFF; 2 concurrent generations; watermark-free image export. Basic for GPT Image 2 and Nano Banana 2 is only available for 1K; 2K/4K uses credits and is settled with package discounts. Example credit costs: Qwen Image 2.1 1K 4.8 credits, 2K 12.8 credits; Nano Banana 2 Lite 4.8 credits; Nano Banana 2 1K 9.6 credits, 2K 19.2 credits.
• Pro (Most Popular): $358.80 per year, automatically renews, cancel at any time. Automatic image generation 60% OFF; other base models 60% OFF; SOTA models 30% OFF; 4 concurrent generations; full commercial rights. Example credit costs: Qwen Image 2.1 1K 3.2 credits, 2K 11.2 credits; Nano Banana 2 Lite 3.2 credits; Nano Banana 2 1K 6.4 credits, 2K 16.8 credits.
• Max: $838.80 per year, automatically renews, cancel at any time. Automatic image generation 1 credit each time; other base models each time 3 credits; SOTA models 50% OFF; 8 concurrent generations; high-volume creation and priority support. Example credit consumption: Qwen Image 2.1 1K 3 credits, 2K 8 credits; Nano Banana 2 Lite 3 credits; Nano Banana 2 1K 3 credits, 2K 12 credits. The page mentions that subscriptions renew monthly or annually, including periodic credits and membership benefits. One-time credit packs do not renew, include commercial usage rights, and provide watermark-free image export when their balance is sufficient to cover the full generation cost; however, extra concurrency, private generation by default, and queue priority still require a subscription. Subscription, registration, and check-in credits each have expiration dates, shown on the credits page. Unused subscription credits do not roll over automatically, while separately purchased one-time credit packs do not expire. Generations on the free plan are public and watermarked by default; new generations for Starter, Pro, and Max members are private and watermark-free by default, and can also be made public. Users currently cannot recover deleted creations.

[Commercial Use and Privacy] While a Starter, Pro, or Max membership is active, generated content may be used for commercial purposes; content generated using purchased one-time credit packs may also be used commercially. The free plan does not include commercial usage rights. Users are responsible for the rights to the materials they upload and for complying with applicable laws and third-party platform rules.

[How to Start] Official instructions state that you can use Google sign-in to get the registration credits currently displayed on the page, and that no credit card is required to start. When not signed in, you can prepare creations, but you must sign in before submitting. The page also offers a "8 free credits" starting option.

[Restrictions] Generation time, available models, and settings vary by task; standard reference image input supports PNG, JPEG, and WebP; users currently cannot recover deleted creations. The page does not provide more details such as support for other formats, specific model parameters, or third-party compliance certifications.