Benchmarks

A dot plot of time-to-first-token across four hardware configurations, one point clearly worse than the rest

Apple's Best MLX Number Lives in a Tab a Fetcher Never Gets

Apple’s MLX post verifies on every load-bearing figure. Its best number sits in a table tab that is not server-rendered, so a fetcher-based reader never sees it. A full recount of all 32 parallelism cells.

A horizontal bar chart of vulnerability categories, two short bars among several long ones

Three Fixes Work, Two Do Not, and SSRF Stays at 2.4%

An agent benchmark where the top performer on functional tests wrote the least secure code. Three interventions measurably help, two do not, and two weakness families stay near zero throughout.

A funnel of particles narrowing through rings, some breaking apart at the exit

CheckerBench: 78% Compiled, 32% Were Accepted

A new benchmark had AI agents write reusable static-analysis checkers. 78.25% compiled, 73.65% told the broken revision from the fixed one, 32.30% were accepted. The loss is almost entirely after detection, and the two obvious explanations both test small.

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.