Podcast Summary: The Last Invention is AI
Episode: 10X Image Dominance: $140M Fal Milestone
Host: Jaden Schaefer
Date: December 30, 2025
Main Theme / Purpose
This episode dives into FAL AI’s recent milestone: raising $140 million in a Series D funding round and unveiling Flex 2 Dev Turbo, a new, dramatically more efficient image generation model. The host analyzes FAL’s technical innovations, broader impacts on the AI industry (especially for developers and business applications), and how this could shape the future of generative media.
Key Discussion Points & Insights
1. FAL’s Funding and Breakthrough Model
- $140M Series D Funding: Signals confidence in FAL’s approach and technology.
- New Model Launch: FAL unveils Flex 2 Dev Turbo, built atop the open-source Flux 2 model by Black Forest Labs.
- “They've just unveiled a flex to dev turbo. This is a faster, cheaper and more efficient version of the open weight model which was originally released by Black Forest Labs.”
(Jaden Schaefer, 03:14)
- “They've just unveiled a flex to dev turbo. This is a faster, cheaper and more efficient version of the open weight model which was originally released by Black Forest Labs.”
2. The Technology: Flux 2 and Lora Adapter
- Base Model: Flux 2—originally open-sourced, powered image generation for Grok on X.
- FAL's Enhancement: Rather than creating an entirely new model, FAL layers a Lora adapter—a lightweight optimization layer—on top of the base model.
- “It's what's called a Lora adapter. This is a lightweight optimization layer that essentially attaches to the original Flex 2 base model and then when they attach it, it dramatically improves the performance.”
(Jaden Schaefer, 04:05)
- “It's what's called a Lora adapter. This is a lightweight optimization layer that essentially attaches to the original Flex 2 base model and then when they attach it, it dramatically improves the performance.”
- Result: 10x cheaper and 6x faster image generation—significant efficiency for developers and businesses.
3. Industry Implications & Open Source Movement
- Repercussions for Major Players: Potential for cheap, fast, open-source image generation to shake up companies like OpenAI and Claude.
- “If you could make this thing six times faster and 10 times cheaper, I think that would be an incredible innovation.”
(Jaden Schaefer, 06:01)
- “If you could make this thing six times faster and 10 times cheaper, I think that would be an incredible innovation.”
- Openness vs. Commercial Use:
- While the model is available for download (via Hugging Face), it is non-commercial unless a separate agreement is made:
“...this model is governed by the Flux non Commercial License V2O which essentially allows personal use research, internal evaluation, but doesn't let you use this for any revenue generated applications.”
(Jaden Schaefer, 20:15)
- While the model is available for download (via Hugging Face), it is non-commercial unless a separate agreement is made:
4. FAL’s Business Model: AI Media Infrastructure, Not Just Models
- Shift from Being a Model Company: FAL positions itself as infrastructure—a hub for real-time generative media across images, video, audio, and 3D.
- “They're really positioning themselves not as a model company, but as an AI media infrastructure...offering developers access to both open and proprietary models...”
(Jaden Schaefer, 08:00)
- “They're really positioning themselves not as a model company, but as an AI media infrastructure...offering developers access to both open and proprietary models...”
- Developer Focus: Over 2 million developers use the FAL platform, with billing based on asset or token usage—enabling flexibility and scalability.
- Supported Use Cases: Ranges from solo indie projects to massive enterprise pipelines for retail, entertainment, and design.
5. Technical Leap: From 50 to 8 Steps
- Distillation Success: Flex 2 Dev Turbo reduces generation steps from 50 (in original) to just 8 per high-quality image—key to efficiency.
- “When the original Flex 2 required roughly 50 inference steps to produce a high, you know, a really high quality image, Turbo achieves a really comparable output in just eight steps. So going from 50 steps to eight steps, this is a massive improvement.”
(Jaden Schaefer, 12:01)
- “When the original Flex 2 required roughly 50 inference steps to produce a high, you know, a really high quality image, Turbo achieves a really comparable output in just eight steps. So going from 50 steps to eight steps, this is a massive improvement.”
6. Benchmark Performance & Cost
- Top Ratings: Turbo scores highest on open-weight image model leaderboards (ELO 1,166 and YUP benchmark 1,024).
- “Turbo now holds the highest ELO score among all of the open weight image models. It has a rating of 1,166, which is passing...competitors like Alibaba on the YUP benchmark.”
(Jaden Schaefer, 13:30)
- “Turbo now holds the highest ELO score among all of the open weight image models. It has a rating of 1,166, which is passing...competitors like Alibaba on the YUP benchmark.”
- Cost-Effectiveness: Generates images for $0.008 each—the lowest cost in current leaderboards.
- “It can create an image for .008 dollars per image. It's basically the lowest price that is currently on the leaderboards.”
(Jaden Schaefer, 14:23)
- “It can create an image for .008 dollars per image. It's basically the lowest price that is currently on the leaderboards.”
7. Downstream Impact & Use Cases
- Real-World Example: AI music platforms like Suno, which generate album covers for every song, represent major demand for scalable, cheap image generation.
- “For example, one of them is Suno AI, which is a music generator. Every single time it creates a song for you, it also creates like an album cover for you....that is millions and millions and millions, perhaps a day.”
(Jaden Schaefer, 15:08)
- “For example, one of them is Suno AI, which is a music generator. Every single time it creates a song for you, it also creates like an album cover for you....that is millions and millions and millions, perhaps a day.”
- Importance for Media-Heavy Workflows: Speed and cost improvements could power a wide range of creative and commercial applications that depend on mass image generation.
Notable Quotes & Memorable Moments
-
On Frustration with Slow Image Generation:
“It really is sort of annoying to sit there and wait for like two minutes for my image to be generated....If you could make this thing six times faster and 10 times cheaper, I think that would be an incredible innovation.”
(Jaden Schaefer, 06:01) -
On FAL’s Developer Ecosystem:
“More than 2 million developers are now using their platform.”
(Jaden Schaefer, 08:30) -
On FAL’s Growth:
“They said apparently on their whole platform in the past year they've become one of the fastest growing backend providers for API generated media. They said they're serving billions of assets each month.”
(Jaden Schaefer, 10:50)
Timestamps for Important Segments
- [03:14] – Flex 2 Dev Turbo Overview & Relation to Flux 2
- [04:05] – Explanation of Lora Adapters
- [06:01] – The Impact of Efficiency Gains
- [08:00] – FAL's "Infrastructure over Models" Strategy
- [12:01] – Technical Achievement: Step Reduction
- [13:30] – Benchmark Scores and Comparisons
- [14:23] – Cost Per Image and Industry Implications
- [15:08] – Image Generation at Scale: Suno AI Example
- [20:15] – Noncommercial Licensing, Access, and Agreements
Overall Tone
Jaden is enthusiastic and practical, highlighting both the technical marvel and the real-world impact of these advances—especially for developers and businesses who stand to benefit from cheaper, faster, and more open image generation AI.
Conclusion
FAL’s latest funding and technical leap demonstrate a pivotal movement in AI image generation: dramatic speed/price improvements, bolstered by open models and innovative engineering, promise to reshape the creative and commercial AI landscape. The episode underscores how infrastructure players like FAL are positioned to transform not just model performance but the economics of generative media at scale.