Bria-2.3-Fast

Version: 1

Bria•Last updated October 2025

Models from Microsoft, Partners, and Community

Models from Microsoft, Partners, and Community models are a select portfolio of curated models both general-purpose and niche models across diverse scenarios by developed by Microsoft teams, partners, and community contributors

Managed by Microsoft: Purchase and manage models directly through Azure with a single license, world class support and enterprise grade Azure infrastructure
Validated by providers: Each model is validated and maintained by its respective provider, with Azure offering integration and deployment guidance.
Innovation and agility: Combines Microsoft research models with rapid, community-driven advancements.
Seamless Azure integration: Standard Azure AI Foundry experience, with support managed by the model provider.
Flexible deployment: Deployable as Managed Compute or Serverless API, based on provider preference.

Learn more about models from Microsoft, Partners, and Community

Key capabilities

About this model

Bria 2.3 Fast is a text-to-image foundation model designed for enterprise-grade performance. It delivers high-quality visual outputs with industry-leading low latency, optimized for the GPU it runs on. Built with a strong commitment to legal compliance and responsible AI practices, it ensures safe and scalable generative image capabilities for commercial use.

Key model capabilities

API Interface: Text-to-Image

Endpoint

POST {base_url}/images/generations Note: Replace {base_url} with your serverless endpoint

Sample Input

{
    "prompt": "A serene mountain landscape during sunset with a clear sky and vibrant colors",
    "size": "1024x1024",
    "num_inference_steps": 10,
    "text_guidance": 2.5,
    "negative_prompt": "stormy weather, dark clouds",
    "image_format": "png",
    "seed": 555
}

Use cases

See Responsible AI for additional considerations for responsible use.

Key use cases

With the largest licensed training set in the world, Bria provides a priority open generative AI platform, enabling enterprises to build high quality, risk free, responsible, visual gen AI solutions.

Out of scope use cases

While the data set includes a wide variety of mediums - photography, art, illustrations and more, it does not contain copyrighted materials, such as fictional characters, logos, trademarks, public figures, harmful content, or privacy-infringing content.

Pricing

Pricing is based on a number of factors, including deployment type and tokens used. See pricing details here.

Technical specs

Training cut-off date

The provider has not supplied this information.

Training time

Time (GPU hours)	Power Consumption (W)	Carbon Emitted (tCO2eq)
36,096	4 MWh	4000 kWh × 0.475 kg CO₂/kWh = 1900 kg CO₂

Input formats

Textual Prompt (Required)
Resolution (Required): Multi-aspect ratio support, approximately 1024x1024 pixels. Size must be one of 1024x1024, 832x1216, 1216x832, 896x1152, 1152x896, 896x1088, 1088x896, 768x1344, 1344x768
Additional Parameters (Optional)
- Negative Prompt
- Number of Diffusion Steps: Must be between 8 and 12
- Guidance Scale: Allows amplification or reduction of the prompt effect. Must be a float between 1.0 and 5.0.
- Image Format: The format of the generated image.
- Seed: The seed for random generation.
NOTE : Include extra-parameters: pass-through in the header

Output formats

Image

Supported languages

The provider has not supplied this information.

Sample JSON response

{
    "data": [
        {
            "b64_json": "<generated_image_in_base64>",
            "nsfw": false
        }
    ],
    "created": 1737976440
}

Model architecture

Bria 2.3 is based on Latent Diffusion Models with a proprietary training dataset of high-quality images for commercial use. Key components include: Text Encoder: CLIP-based Perceptual Encoder-Decoder: VAE Backbone: UNet Images are transformed into latent representations through the VAE, and textual captions are encoded using the text encoder. The model supports a resolution of 1024x1024 with multi-aspect ratio generation, curated for aesthetic quality.

Long context

The provider has not supplied this information.

Optimizing model performance

Bria 2.3 offers the best balance between quality and speed. Other versions include Bria 2.3 HD (high-quality output) and Bria 2.3 Fast (reduced latency).

Additional assets

The provider has not supplied this information.

Training disclosure

Training, testing and validation

Bria 2.3 fast is trained on the largest licenced data set in the world - ~1B images from leading Image providers including Getty Images, Envato, Depositphotos, Alamy, Freepick and more. While the data set includes a wide variety of mediums - photography, art, illustrations and more, it does not contain copyrighted materials, such as fictional characters, logos, trademarks, public figures, harmful content, or privacy-infringing content.

Distribution

Distribution channels

As an open platform - You can finetune Bria's models and host your solutions according to your data privacy policies.

More information

With a clear agenda of practicing Gen AI while respecting the data owners and creatives, Bria has developed a patented attribution agent for rewarding data owners for generations contributed by their data. Bria AI Responsible AI Policy : In this document, we share the core fundamentals that guide every decision we make at Bria. These principles are the backbone of our commitment to the highest standards of responsible AI practice and reflect our dedication to leading in the development of accountable AI technologies. We invite all stakeholders to join us on this journey towards accountable AI innovation. Bria's Attribution Technology: Advancing Equitable AI Development Bria's attribution technology is a comprehensive system designed to track, attribute, and fairly compensate for content used in AI model training. This technology represents a significant leap forward in addressing key ethical and legal challenges in AI development. Contact for Questions Support@bria.ai

Responsible AI considerations

Safety techniques

Built with a strong commitment to legal compliance and responsible AI practices, it ensures safe and scalable generative image capabilities for commercial use. While the data set includes a wide variety of mediums - photography, art, illustrations and more, it does not contain copyrighted materials, such as fictional characters, logos, trademarks, public figures, harmful content, or privacy-infringing content. With a clear agenda of practicing Gen AI while respecting the data owners and creatives, Bria has developed a patented attribution agent for rewarding data owners for generations contributed by their data.

Safety evaluations

The provider has not supplied this information.

Known limitations

The provider has not supplied this information.

Acceptable use

Acceptable use policy

Bria AI Responsible AI Policy : In this document, we share the core fundamentals that guide every decision we make at Bria. These principles are the backbone of our commitment to the highest standards of responsible AI practice and reflect our dedication to leading in the development of accountable AI technologies. We invite all stakeholders to join us on this journey towards accountable AI innovation. Bria's Attribution Technology: Advancing Equitable AI Development Bria's attribution technology is a comprehensive system designed to track, attribute, and fairly compensate for content used in AI model training. This technology represents a significant leap forward in addressing key ethical and legal challenges in AI development.

Quality and performance evaluations

Source: Bria The provider has not supplied this information.

Benchmarking methodology

Source: Bria The provider has not supplied this information.

Public data summary

Source: Bria The provider has not supplied this information.

Model Specifications

LicenseCustom

Last UpdatedOctober 2025

Input TypeText

Output TypeText,Image

ProviderBria

Languages1 Language

Quick Start