Blog
Narrative posts that connect releases and serving concepts back to canonical reference pages.
- How diffusion generation evolved from pixel U-Nets to transformers, flow matching, and modern video models
A narrative arc from denoising diffusion probabilistic models and convolutional U-Net backbones through CLIP-conditioned image systems, latent diffusion, diffusion transformers, flow matching, and source-backed open video and world-model examples.
- Foundations
- Model Family
- LLMs are no longer wholly reliant on the internet
Why modern language-model quality still starts with internet-scale pretraining but increasingly depends on mid-training, post-training, preference feedback, verifiable rewards, and distillation loops.
- Foundations
- Alignment
- the best computer for local language models (2026)
An overall guide to the best computer to buy for local language models. We recommend an M-series laptop or a 5090.
- Inference
- Local Models