My Qwen3.8 27B benchmarks on a 16GB RTX 5070 Ti: base vs. Swift 1.5 for coding, Bonsai 2 for speed, with tested settings you can try.

Qwen3.8 27B on 16GB VRAM: RTX 5070 Ti benchmarks

Explore Xiaomi's MiMo-V2.6-Pro in an interactive 3D animation: how tokens, attention, a mixture of experts and speculative decoding produce an answer.

First Light: MiMo-V2.6-Pro's architecture in 3D

Take a night drive through Kimi K3 as an interactive 3D city: follow one token through Kimi Delta Attention, MLA, attention residuals and 896 experts.

Neon City: Kimi K3's architecture in 3D

Take the red pill and follow one token through DeepSeek V4-Pro in interactive 3D: compressed attention, hyper-connections, 384 experts and 61 layers.

Follow the White Rabbit: DeepSeek V4's architecture in 3D

Press start and watch a large language model finish a sentence in an interactive visualization: tokens, attention, the KV cache and mixture of experts.

Insert Token: how LLMs work, one token at a time

How a machine that only predicts the next word ends up building world models, planning ahead, and doing things that look like thinking.

How AI models think

Six principles for using AI effectively: prepare the context, aim at goals instead of tasks, master the handoff, match the model to the task, let the AI triage your decisions, and tinker.

Principles for using AI effectively

Save time setting up Sunshine on Windows 11 with this helpful guide.

How to setup Sunshine for Steam Deck streaming

Windows can be a bloated mess, but with a little work, it can be turned into a delighful OS.

How to setup Windows 11

Explain how you understand language and what are your core competencies as an AI model.

Blog

Subscribe to Blog

All Posts

Always bet on JavaScript

Indie Maker Syndicate Website

Hello World