Bria needed to bring new models to market faster while expanding developer adoption
For AI model developers, building a great model is only part of the challenge. Getting it into the hands of developers and enterprises quickly, reliably, and at scale is equally important.
Before fal
Bria continuously develops new visual AI capabilities, from its flagship FIBO generation models to specialized editing tools. Each release creates an opportunity to reach new customers, but also introduces operational and distribution challenges.
While Bria already operates its own production-grade inference infrastructure for enterprise customers, expanding its public model catalog required additional flexibility. Managing GPU capacity, deploying new endpoints, and scaling to unpredictable demand could divert engineering resources away from model development.
Equally important was distribution. Bria needed a way to introduce each new release to developers actively building generative AI applications, without having to build that audience entirely from scratch.
Why fal
The Bria team looked for a partner that could:
- Enable engineers to deploy and update models independently, without lengthy infrastructure setup or external release dependencies.
- Handle growing production workloads and traffic spikes without requiring significant over-provisioning.
- Make new models easily discoverable and accessible to a large community of developers and enterprises.
- Collaborate on model launches, marketing, and customer acquisition beyond providing infrastructure alone.
“fal doesn't act like an infrastructure vendor; it acts like a go-to-market partner. Its partnership, marketing and sales teams invest in the models on the platform and help put them in front of the right audiences. For a company like Bria, that reach matters as much as the compute.”
Misha Feinstein, CTO, Bria
Building with fal
Bria partnered with fal in 2024 to make its models accessible to developers through fal's inference infrastructure and model marketplace.
As its catalog expanded, Bria adopted fal Serverless to independently deploy and manage its own models. Today, Bria's engineers can take a model from research-ready to publicly available in days rather than weeks, without managing the underlying GPU infrastructure.
Through fal, Bria benefits from:
- The ability to deploy, configure, and update models independently using the fal SDK, giving engineers full control over release schedules, endpoints, and playground experiences.
- Automatically scaling GPU capacity that supports production workloads and traffic spikes without requiring Bria to provision infrastructure for peak demand.
- Access to fal's developer ecosystem through ready-to-use APIs, interactive playgrounds, and documentation that make Bria's models easy to discover, test, and integrate.
- Joint launch announcements, marketing campaigns, and sales support that help introduce each new model to relevant developers and enterprise customers.
This partnership has supported three major launches in ten months: FIBO in October 2025, FIBO Edit in January 2026, and FIBO 1.5 in August 2026.
With infrastructure and distribution handled through fal, Bria can focus more engineering resources on developing and improving its models while maintaining its own infrastructure for direct enterprise customers.
“The moment we ship something new, it's available to fal's community, and with fal's marketing behind it, it reaches people fast. Day to day, that means our engineers stay focused on building and improving Bria's models instead of managing serving infrastructure.”
Misha Feinstein, CTO, Bria
Results
- Bria's models served tens of millions of generations through fal in 2026, supporting developers building production applications across a wide range of visual AI use cases.
- Tens of thousands of distinct fal customers used Bria's models in 2026, giving Bria access to an established developer ecosystem without needing to acquire every customer directly.
- Monthly users of Bria's models on fal grew nearly threefold from early summer to a record high in September 2026, following the launch of FIBO 1.5.
- Hundreds of fal customers with active enterprise contracts used Bria's models, accounting for approximately one-third of Bria's generation volume on the platform.
- Bria reduced the time required to bring research-ready models to developers to days rather than weeks, with hundreds of self-serve deployments across dozens of applications in 2026.
- fal also helps introduce Bria to developers and businesses that can grow into larger production deployments through Bria's direct enterprise offering.
The partnership demonstrates how model developers can combine scalable inference infrastructure with an established developer ecosystem to accelerate both product launches and commercial adoption.
“Infrastructure can be replaced; a partner that actively helps grow your business is much harder to find. That's why we're doubling down on the relationship and on building more together.”
Misha Feinstein, CTO, Bria
