DreamStudio is an AI image generation and editing platform launched by Stability AI. Based on the Stable Diffusion series of models, it provides text-based images, image-based images, fine editing and branded creative production functions, covering both web applications and API delivery forms. In 2026, it has evolved into the platform Brand Studio, oriented to enterprise-level brand creative production.

DreamStudio

Core parameters and statistics of DreamStudio

DreamStudio (now upgraded to Brand Studio) is Stability AI's AI image generation and editing platform for creative production. It is not a single model product, but a multi-layered product system built around the Stable Diffusion series of models: the front-end has a web application (Brand Studio), the back-end has a pay-as-you-go developer API (Stability AI Developer Platform), and the middle layer also has enterprise-oriented brand customization solutions.

Projects	Public Information
Full product name	DreamStudio → Brand Studio (2026-04 upgrade)
Affiliated Company	Stability AI Ltd.
Headquarters	London, United Kingdom (GB)
Product Positioning	Branded Creative Production Platform (Creative Production Platform)
Underlying model	Stable Diffusion 3.5 Large / Turbo / Medium / Flash, SDXL 1.0, Stable Image Ultra / Core
Core capabilities	Inpaint, Inpaint, Outpaint, structure control, style transfer, super resolution, background replacement 3D generation, audio generation
Delivery form	Brand Studio (Web application), REST API, self-hosted (Self-Hosted License)
Enterprise Customers	Arm, AWS, Microsoft Azure, NVIDIA, HubSpot, Electronic Arts, UMG, Warner Music Group
Community channels	Discord (stablediffusion), GitHub, Hugging Face

Product evolution: DreamStudio was originally launched in 2022 with the open source release of Stable Diffusion, and is positioned as a demonstration and experimental portal for text-to-image generation. In April 2026, Stability AI officially upgraded it to Brand Studio, shifting from "AI image tool" to "branded creative production platform", adding enterprise-level capabilities such as Brand Central (brand identity management center), Producer Mode (multi-step workflow orchestration), and Curated Model Routing (cross-model intelligent routing). This evolution means that its target customer base has expanded from individual creators to brand marketing teams and creative agencies.

Model Matrix: Brand Studio does not only run Stability AI self-developed models. The Curated Model Routing layer automatically selects the most appropriate model based on the type of task (product photography, concept art, background replacement), including third-party models. The API level provides a complete ladder from SDXL 1.0 (0.9 points/time) to Stable Image Ultra (8 points/time).

User and market recognition

Market recognition of DreamStudio/Brand Studio is reflected more in enterprise-level customer endorsements and industry influence than in public end-user numbers (the latter is not disclosed).

Enterprise Customer Matrix: The Stability AI official website discloses leading enterprise customers such as Arm, AWS, Lenovo, Microsoft Azure, NVIDIA, HubSpot, Electronic Arts, Universal Music Group, and Warner Music Group. This customer structure shows that its products have passed the individual/small team experimental stage and entered the purchase list of brand enterprises and large platforms.

Industry case quantification: Three public cases provide verifiable effect data - HubSpot used Stability AI tools to achieve "instant generation of brand visual materials" on Amazon Bedrock; Mercado Libre (the largest e-commerce company in Latin America) achieved a 25% higher click conversion rate through AI-generated product images; online education platform Stride Learning launched a personalized story generation application within 6 months, and can produce 1,000+ pictures per minute. These data directly prove the ROI in marketing content production and e-commerce scenarios.

Community Ecology: The Stable Diffusion open source community is one of the largest moats for Stability AI. Around the SD series models, a huge user base and third-party tool ecosystem (ComfyUI, Automatic1111, etc.) have been accumulated on Discord, Hugging Face, and GitHub. However, it should be noted that community popularity is not equal to the paid conversion rate of Brand Studio products, and there is an obvious funnel between the two.

Cost Advantages of DreamStudio

The cost advantage is the core difference between DreamStudio/Brand Studio and competing products such as Midjourney and DALL-E, but its "cheapness" is reflected in architectural flexibility and pay-as-you-go, rather than simply absolute low price.

C-side/individual creator: Brand Studio provides a free trial plan (first 1,000 points), and the Core plan is $50/month (5,000 points, points are not accumulated). Compared with Midjourney’s $10-60/month fixed subscription system, Brand Studio’s points system is more friendly to low-frequency users (purchase on demand after a free trial), but may be more expensive for high-frequency users – 5,000 points is equivalent to approximately 625 Stable Image Ultra generation or 2,000 SDXL generation under the Core plan. High-frequency creators may need to purchase additional points packages.

Developer/API Call: Stability AI API uses a points system, 1 point = $0.01. The core generative model price ladder is as follows:

Model	Unit price (points/time)	Equivalent in US dollars	Typical uses
Stable Image Ultra	8	$0.08	Highest quality flagship output
SD 3.5 Large	6.5	$0.065	High Quality Universal Build
SD 3.5 Large Turbo	4	$0.04	Fast, high-quality output
SD 3.5 Medium	3.5	$0.035	Balancing quality and speed
Stable Image Core	3	$0.03	Economical universal generation
SD 3.5 Flash	2.5	$0.025	High throughput and rapid generation
SDXL 1.0	0.9	$0.009	Low-cost, stable output

Editing operations (Inpaint, Erase, Search & Replace, etc.) are unified to 4-5 points/time, background replacement and relighting are 8 points/time, and super scores range from 2 (Fast) to 60 (Creative) points/time. Comparing the DALL-E 3 API (approximately $0.04-0.08/image) and Midjourney’s no public API, Stability AI’s API is significantly richer in model selection flexibility and cost gradient, but the unit price is not superior to the flagship model.

Enterprise/Private: Brand Studio Enterprise is a customized quotation (Custom), including unlimited seats, custom points package Brand Central Producer Mode, SSO and project-level permission control. There is also a Self-Hosted License path, which is suitable for organizations with high data sovereignty requirements - the subscription fee is converted into infrastructure and maintenance costs, but complete data isolation and model customization capabilities are available.

Main features of DreamStudio

The functions of DreamStudio/Brand Studio are organized around a "complete pipeline from prompt to brand-level finished product" rather than a single Vincent diagram tool. The following are the five core competency groups:

• Vincent Diagram and Multi-Model Routing: Supports image generation through natural language description, and the underlying layer automatically matches Stable Image Ultra, SD 3.5 Large/Turbo/Medium/Flash, SDXL and other models through Curated Model Routing. Hidden linkage: The routing layer not only selects models based on quality, but also recommends the best model based on task semantics (product photos vs concept art vs realistic portraits) to reduce user trial and error costs.

• Fine editing and structure control: including Inpaint (mask replacement), Outpaint (expand the canvas outward), Erase Object (erase object), Remove Background (remove background), Search & Recolor/Search & Replace (semantic search and replacement), Replace Background & Relight (background replacement + heavy light). Expert View: The greatest value of this set of functions is that "no precise masking is required" - Search & Replace allows describing "replace a red car with a blue one" in natural language, which lowers the operating threshold by an order of magnitude compared to traditional Photoshop's pixel-by-pixel operation.

• Image Super-Resolution and Restoration: Creative Upscaler (4K upscaling with prompt guidance), Conservative Upscaler (lossless 4K upscaling), Fast Upscaler (fast 4x upscaling). The three cover the entire link from "quick preview" to "refined output", directly responding to the core complaint about the insufficient resolution of AI-generated images.

• Style control and brand consistency: Structure (structure reference), Sketch (line draft/sketch guide), Style Guide (style reference), Style Transfer (style transfer). Hidden linkage: Combined with Brand Central's Brand ID model, the team can "train" the brand color palette, visual style, and photography specifications into the model, so that each generation abides by the brand guidelines by default, solving the implementation pain point of "AI images are good-looking, but they don't look like our brand."

• 3D and Audio Generation (Platform Extension): Stable Fast 3D (single image to 3D asset, 10 points), SPAR3D (real-time 3D editing, 4 points), Stable Audio 2.5/3.0 (text/audio to music, 20-26 points). These capabilities illustrate that Stability AI's product strategy is a "multi-modal creative middle platform" and is not limited to image generation.

Model and version evolution of DreamStudio

DreamStudio's product version iterations can be tracked from two dimensions—the upper product form (Web application) and the underlying model capabilities (inference engine).

Main line of product form

• Brand Studio (2026-04-08): A branded creative production platform upgraded from DreamStudio. Added Brand Central (brand identity training and management), Producer Mode (multi-step workflow), Curated Model Routing (intelligent routing across models), team collaboration and SSO. This is the most important leap in product positioning.
• DreamStudio Web (~2022-08): The initial release version, launched with the open source release of Stable Diffusion, provides basic text-generating and graph-generating capabilities. The early version attracted users with a free points system and accumulated early awareness.
• DreamStudio API (continuous iteration): The developer platform continues to update the model list and endpoints. In 2026, it will cover the six API categories of Image, Upscale, Edit, Control, 3D, and Audio. The model name has evolved from the earliest SD 1.4/1.5 to the current SD 3.5 series and Stable Image Ultra.

Underlying model version context

Stability AI’s model iteration is the true engine of DreamStudio’s product capabilities:

Model nodes	Release time	Key changes
Stable Diffusion 1.4/1.5	2022-08	Starting point for open source image generation, 512x512 output
SDXL 1.0	2023-07	1024x1024 native resolution, quality jump
SD 3.5 Large (8B)	2024-10	Maximum parameter basic model, prompt word compliance is significantly improved
SD 3.5 Large Turbo	2024-10	Turbo distilled version, inference steps halved
SD 3.5 Medium (2.5B)	2024-10	Efficient models runnable on consumer-grade GPUs
SD 3.5 Flash	2025	Extremely fast distilled version, suitable for high throughput scenarios
Stable Image Ultra	2026	Flagship imaging service, highest quality and detail
Stable Audio 3.0	2026-05	Up to 6 minutes of high-quality audio generation

Technical advantages of DreamStudio

The technical advantages of DreamStudio/Brand Studio come from the collaboration of the three-tier architecture, rather than the performance leadership of a single model.

Architecture layering: The bottom layer is the open source model family (SD 3.5, SDXL, Stable Audio, Stable 3D) independently developed by Stability AI, the middle layer is the Curated Model Routing intelligent routing engine, and the top layer is the Brand Studio application for brand teams. The actual effect of this layering is that enterprises do not care which model is used at the bottom layer. The routing layer automatically selects the optimal model based on task semantics, reducing the selection cost for non-technical users.

Open Source Ecological Barriers: The core strategy of Stability AI is "open source core model + closed source enterprise layer". The open source route brings three advantages - community-contributed fine-tuning models and LoRA greatly enrich the ecosystem, third-party tool chains (ComfyUI, Automatic1111, etc.) lower the user acquisition threshold, and academics and enterprises can fine-tune models on their own data. These constitute ecological network effects that are difficult for latecomers to replicate in the short term.

Brand ID fine-tuning: Brand Central supports training customized Brand ID models, encoding brand visual specifications directly into the generation process. This solves the biggest pain point when enterprises adopt AI image tools - "AI-generated images have random styles and cannot guarantee brand consistency." Compared with Midjourney's Style Reference and DALL-E's simple style presets, Brand ID's depth of customization and binding of brand elements are significantly higher.

Full-modal expansion route: Stability AI is not only deployed in the image field. Stable Audio 2.5/3.0, Stable Fast 3D, and SPAR3D indicate that it is building a full-modal creative tool matrix of "image + audio + 3D". For the brand team, this means that in the future, the production of visual and audio 3D materials can be completed within the same platform, reducing the context loss of switching between multiple tools.

How to use DreamStudio

DreamStudio/Brand Studio provides three usage paths, with different entrances, costs and applicable scenarios:

How to use	Entrance	Suitable for the crowd	Cost	Key constraints
Brand Studio Web	brandstudio.com	Brand teams, creative agencies, individual creators	Free trial (1000 points) → Core $50/month (5000 points) → Enterprise customization	Core plan points do not accumulate; Brand ID and Producer Mode are only for Enterprise
REST API	platform.stability.ai	Developer SaaS products, automated workflow	Pay-as-you-go, points system (1 point = $0.01), minimum 0.9 points/time	An account needs to be created and recharged; frequency control and concurrency limits are subject to the official real-time page
Self-hosted License	stability.ai/license	Enterprises/institutions with high data sovereignty requirements	Business quotation	Need their own GPU infrastructure and operation and maintenance capabilities

Typical usage steps (API):

1. Visit platform.stability.ai to register an account and create an API Key.
2. Top up points (minimum $10, get 25 free points).
3. Call the generation endpoint: select the model (such as stable-image/ultra), and pass in prompt, aspect_ratio, output_format and other parameters.
4. Receive the returned image Base64 or download it directly, and the corresponding points will be deducted on a per-time basis.

# API call example (Vincent diagram)
curl -X POST https://api.stability.ai/v2beta/stable-image/generate/ultra \
  -H "authorization: Bearer <YOUR_API_KEY>" \
  -H "accept: image/*" \
  -F "prompt=cinematic product photo of a leather bag on marble surface, studio lighting" \
  -F "aspect_ratio=16:9" \
  -F "style_preset=photographic" \
  -F "output_format=jpeg"

Product Pricing for DreamStudio

The pricing system of DreamStudio/Brand Studio is divided into three tiers, covering the complete path from personal trial to enterprise-level deployment.

Brand Studio Subscription Plan:

• Trial (free): Includes the first 1,000 free points, including all functions of the Core plan (Curated Model Routing, Vincentian diagrams, and editing Precision Tools). You need to upgrade after the points are used up. Suitable for individual creators and teams for initial evaluation.
• Core ($50/month): 5,000 points per month, includes a full set of generation and editing tools. Points are issued on a monthly basis and will not be accumulated if they are not used up in the current month. Ideal for creative professionals and small teams.
• Enterprise (customized quotation): Unlimited seats, custom points package Brand Central (brand ID training and hub), Producer Mode (multi-step workflow orchestration), SSO and project-level permission control, white-glove support. Suitable for brand teams and medium to large enterprises.

API metered billing: 1 point = $0.01. Core image generation ranges from SDXL 1.0 (0.9 credits) to Stable Image Ultra (8 credits); Editing 4-8 credits; Ultra 2-60 credits; 3D 4-10 credits; Audio 20-26 credits. For detailed price matrix, see the table in the "Cost Advantages" chapter.

Explicit Cost vs. Implicit Cost: For API users, the real total cost is not just the consumption of points. Developers need to take the following factors into their budget - the cost of trial and error in the prompt debugging phase (inefficient prompts lead to multiple generations), improper model selection (use Ultra to complete Flash-level tasks), and frequency control limitations in high-concurrency scenarios (business communication may be required to increase the quota). In the Enterprise plan, Brand ID model training and Producer Mode workflow construction also require labor investment. These are hidden costs that are not shown on the pricing page.

Application scenarios of DreamStudio

The capability distribution of DreamStudio/Brand Studio determines that it is suitable for the two core scenarios of "visual content mass production" and "branded creative delivery", rather than exploratory artistic creation.

• E-commerce product visual production: Brand e-commerce teams can use Structure control (refer to product image structure), Search & Replace (replace background/color), Replace Background & Relight (scenario-based scenery) to generate product display images in batches. Quantitative deduction: An e-commerce operation needs to produce 50 product scene pictures every week - the traditional process (photography + photo editing) takes about 8-10 hours; the Brand Studio pipeline can compress the time to 1.5-2 hours, without the need for outdoor photography and models. Mercado Libre’s 25% CTR improvement has validated the ROI of this scenario.

• Brand Social Media Content Batch: Combined with Brand Central's brand ID model, the marketing team can batch generate dozens of social media graphics, advertising banners, and infographics that comply with brand visual specifications every week. Human-computer collaboration boundary: The first draft generation can be 100% automated, but the brand visual specification review (color accuracy, logo placement, font compliance) must set up manual confirmation points and cannot be published directly.

• Game and Entertainment Concept Design: A collaboration between Electronic Arts and Stability AI shows that game development teams can accelerate concept design iterations with Sketch control (line draft → fine draft) and Style Transfer (style reference). Cost reduction and efficiency improvement: The traditional first draft cycle of a senior concept designer is about 3-5 days/piece; the AI-assisted pipeline can compress the first draft output to 0.5-1 day/piece, but the refinement and artistic direction control of the final draft still need to be led by the designer.

• 3D Asset Rapid Prototyping: Stable Fast 3D and SPAR3D allow 3D models to be generated from a single image or edited in real time, suitable for product prototype display, e-commerce 3D preview and AR/VR material preparation. The current limitation is that the output accuracy and topological controllability are still far behind professional DCC tools (Blender, Maya), which are more suitable for early concepts and preview scenes.

Applicable groups of DreamStudio

The three-tier product form of DreamStudio/Brand Studio corresponds to different user groups and usage boundaries:

• Brand Marketing and Creative Team: Brand Studio’s core target users. Need to produce high-quality, brand-consistent visual materials in batches, value Brand Central, Producer Mode, and team collaboration features. Prerequisite: The team needs to have clear brand visual specifications (color palette, fonts, photography style), otherwise the training value of Brand ID will be greatly reduced.

• Independent creators and freelance designers: You can experience AI-assisted creation through the Trial or Core plan. Compared with Midjourney's strongly stylized output, Brand Studio is more flexible in editing control and structural reference, and is more suitable for illustrators and graphic designers who need precise control over output content. Not suitable for boundaries: If the creative style relies heavily on hand-made brushstrokes and completely random exploration, Brand Studio's industrial pipeline will limit the creative freedom.

• Developers and AI application integrators: embed image generation capabilities into their own products (such as design tools, e-commerce platforms, content management systems) through REST APIs. The point-based pay-as-you-go system is friendly to integrators, but attention needs to be paid to the frequency control limitations of the API and the changes in output consistency caused by model updates. Recommendation: The production environment should fix the API model version, and verify the output compatibility of the new model in the staging environment before switching.

• Enterprise Buyers and IT Decision Makers: SSO, audit logs, data isolation and Brand Central for Enterprise plans are compliance critical. Not Fit Boundaries: If your organization's image generation needs are sporadic and low-frequency, Enterprise's annual commitment may not be as economical as a per-volume API. Additionally, Stability AI’s financial stability (profitability under an open source business model) should be considered as part of the procurement risk assessment.

Summary and Outlook

The core value of DreamStudio/Brand Studio is to encapsulate the technical capabilities of the Stable Diffusion series of models into a creative production platform for brand teams. It is not the cheapest (compared to SDXL Community Edition self-hosted) nor the easiest to use (compared to Midjourney's Discord out-of-the-box) image tool, but it forms a differentiated positioning in the three dimensions of "brand consistency + fine control + multi-modal extension".

Current limitations: Brand Studio's branding functions (Brand ID, Producer Mode) are locked in the Enterprise plan, and small and medium-sized teams cannot purchase them separately; the API points system is cost-effective for occasional low-frequency scenarios, but the monthly fee in high-frequency batch scenarios may be higher than Midjourney's fixed subscription; although the self-hosted path has obvious data security advantages, the requirements for operation and maintenance capabilities limit its popularity.

Procurement/Adoption Risk Assessment: Three dimensions need to be verified before making a decision - ① The current point consumption rate and the commercial terms of the point package required for future expansion (whether there is a minimum annual commitment); ② The training data usage boundary of the Brand ID model (whether it will be used for secondary training of the Stability AI model); ③ Stability AI since 2026 The product direction is transforming from "open source community" to "enterprise brand platform". Will the release rhythm of the open source model and the community version be weakened - this poses a potential route risk to companies that rely on the community version. It is recommended to first run through the end-to-end process on 1-2 real business scenarios through the trial solution, and then make long-term purchasing decisions after verifying the output quality and manual intervention ratio.

Related tools: midjourney, stable-diffusion

Version Info

• Brand Studio (formerly DreamStudio) ：In April 2026, it was officially upgraded from DreamStudio to Brand Studio, shifting its positioning from an image generation tool to a branded creative production platform, and adding enterprise-level functions such as Brand Central, Producer Mode, and Curated Model Routing. （2026-04-08）
• DreamStudio Web (Classic) ：The initial release version of DreamStudio provides text-to-image generation functions based on Stable Diffusion and is a benchmark product in the field of early AI image generation. There is no official precise date yet. （~2022-08）