Fal.ai at $95M/year growing 4,650% YoY

TL;DR: As generative media took off in 2022-2023, Fal.ai pivoted to an inference-as-a-service API for image, video, and audio models that optimized for speed and ease of use over raw compute. Sacra estimates Fal.ai hit a $95M revenue run rate in July 2025, up from $25M at the end of 2024. For more, check out our full report and dataset on Fal.ai.

Key points via Sacra AI:

As open-source text-to-image models like Stable Diffusion took off in 2022, building generative AI apps on top of them required either self-hosting models on AWS/GCP/Azure or accepting slow, rate-limited public endpoints—inspiring Fal.ai (2021) to launch an inference-as-a-service provider optimized for speed and reliability, serving first Stable Diffusion and now 600+ image, video, audio, and 3D models over an API monetized on usage. At ~$0.03 per image or ~$0.10 per second of video API invocation, developers tend to (1) discover & demo open-source models from Stability AI (Stable Diffusion) or Black Forest Labs (Flux) on Hugging Face and play around with them locally, then (2) set up Fal.ai and Replicate to prototype and start using them in production, and then (3) as usage starts to scale, deploy them on rented bare-metal GPUs via CoreWeave, Lambda Labs, or Crusoe.
After signing enterprise deals to power generative image & video experiences in Quora’s Poe, Canva and Perplexity, Sacra estimates that Fal.ai hit a revenue run rate of $95M in July 2025, up from $25M at the end of 2024 and up 4,650% YoY from about $2M in July 2024, valued at $500M as of its $49M July Series B for a 5.3x forward revenue multiple. Compare to AI neocloud Together AI at $130M ARR at the end of 2024, valued at $1.25B for a 9.6x multiple, “GitHub for models” Hugging Face at $70M revenue in 2023 growing 367% YoY, valued at $4.5B for a 64.3x multiple, OpenRouter at $5M in annualized revenue in May 2025, up from $1M at the end of 2024, valued at $500M for a 100x multiple, and Stability AI at ~$50M revenue in 2024, up 525% YoY, valued at $1B in 2022 for a 666x multiple on $1.5M in revenue.
Because image- and video-generation models are fine-tuned ~1,000x as often as LLMs and usually run as chained workflows (remove background, upscale, recolor) to achieve brand- or character-level consistency that one-shot prompting can’t deliver, Fal.ai is evolving past raw inference & model hosting into workflows, bundling together model chaining, LoRA fine-tuning, and asset storage into one billable integration. Like OpenRouter, Fal.ai offers developers a single integration point to call a wide range of models—not integrating directly with OpenAI’s API is akin to not integrating directly with Stability AI’s API—creating natural expansion paths through added endpoints while benefiting from a flywheel where the more developers & enterprises on Fal.ai, the more valuable it becomes as a distribution channel for model providers.

For more, check out this other research from our platform:

Fal.ai at $95M/year growing 4,650% YoY

Read more from

Fal.ai

Fal.ai revenue, growth, and valuation

Read more from
#ai

Replit at $106M ARR

OpenEvidence at $50M/year growing 30% MoM

OpenEvidence revenue, growth, and valuation

Read more from
#cloud-gpus

Unmentioned Stargate partner

Fluidstack at $180M ARR

Fluidstack revenue, growth, and valuation

Create a free account, or log in.

Free article limit reached.

Standard membership required.