Creative Analytics Agent - identifies top-performing creative elements and spots performance trends
Creative Generator Agent - generates finished ads and strategic creative briefs from winning patterns
AI-powered test ideas from performance data
Meta Ads MCP integration with ChatGPT, Claude, and Gemini
Ad fatigue detection and alerts
Generate variations from winning ads
Pyramid-JiT (P-JiT) architecture with 32×32 token reduction
4× more aggressive token reduction than Linum v2
Converges in 11× fewer samples than Linum v2
2B-parameter model with frozen Qwen text encoder
Pixel-space diffusion with multi-resolution readouts
PixelREPA alignment with DINOv3-L
Curriculum learning over noise schedule
What makes it different
Analyzes performance data to identify exact creative elements that drive results, saving team hours of manual analysis
Reduces creative planning time significantly - one case from 2 days to 30 minutes, another from 1 day to minutes, another from 3-4 hours per client to 20 minutes
Integrates with favorite AI assistants via Meta Ads MCP without leaving ChatGPT, Claude, or Gemini
Generates multiple test-ready ad variations automatically from winning patterns
Low-cost training: full 2B model trained for just $21K in compute
Controls baked into model training rather than post-hoc fine-tuning
Pixel-space compression without separate VAE bottleneck
Apache 2.0 open release with full model weights and code
Automads
Linum