AI Models

Two spheres of different size facing each other across a measuring scale whose gradations are crowded on one side and stretched on the other

Laya vs Jev: A Challenger With Its Own Scorecard

An open-source challenger to Jev ships with a comparison table. The accuracy claim does not survive reading the rest of the page, but the structural argument does.

Four illuminated spheres on a baseline; the rightmost dissolves into particles beside a dashed line continuing into empty space

GPT-6 Sol and Luna: Reading the Benchmark Footnotes

OpenAI’s GPT-6 Sol and Luna announcement is a cost-efficiency pitch. Its own footnotes concede that the closest competitor’s cost is the one figure not reported.

Three small tools branching from a single typed decision node

What People Are Actually Building on Jev

A week after launch is when the real question gets answered: does anyone build on it. For Jev the early answer is code review, model routing and linting - places where which one matters more than write me something.

A typed decision returning a probability distribution rather than a sentence

Jev and TypeSafe AI's System One Models

Every model you have used generates text one token at a time, even when all you needed was yes or no. TypeSafe AI built one that returns typed values instead.

A large model compressed into a fraction of its original memory footprint

Ternary-Bonsai-2-27B: A 27B Model in 6GB

A 27B model squeezed into 5.95GB at 1.72 bits per weight, claiming 98.2% of full-precision performance on its own benchmark suite - and what that compression actually costs.

A generated object with a genuinely transparent checkerboard background, no cut-out edge

Qwen-Image-2.1: Native Transparency and a License Catch

Alibaba built transparency into the image format itself rather than removing backgrounds afterwards - then shipped it under a research licence instead of the Apache 2.0 its predecessors used.

A compressed memory cache shrinking beside an unchanged context window

DeepSeek-V4.1-Flash and the KV Cache Problem

DeepSeek-V4.1-Flash is bigger than its predecessor and cheaper to run. The savings come from cutting KV cache to 890 bytes per token - about a quarter of the previous model.