TST LAB / TOPIC DIRECTORY
Technology & AI
Technical explainers, AI analysis and technology reviews.
← Back to the editorial directory06Published work

NOTES & ESSAYS
A Benchmark Is a Question, Not a Crown
Benchmarks can make model comparisons useful—but task choice, contamination, scoring and judging shape what a result actually means.
Read ↗
AI
Claude Opus 5.5: What the Public Evidence Shows—and Doesn't
A source-based look at Anthropic’s claims, METR’s bounded task evaluation and the editorial process behind its report—and what neither can establish about broad real-world capability.
Read ↗