SF builds its own model benchmarks while everyone else still asks what an agent is
dbreunig · x · 2026-11-09
A wry one-liner from dbreunig capturing the widening gap between the SF AI frontier — where the consensus is that you must build your own benchmarks to properly assess models — and everyone else, who is still asking "What's an agent?"
More from Fun
- AI-generated hills look hand-painted — but the model forgot to add a single tree — misovalko · 2026-11-09
- Opus 5.5 generates instant light/dark mode UIs that PMs would quote 2 months for — cyrus_zei · 2026-11-09
- Alexandr Wang teases Muse Voice: 'your girlfriend is already jealous' — alexandr_wang · 2026-11-09
- 'Software is self-expression now': creator shows 40+ quirky AI-built side projects — yungcontent · 2026-11-09
- Right Answer, Wrong Reasons: GPT-OSS-120B Forecasts Costco Profits Within 4.4% — AI Engineer · 2026-11-09
- Meme: the AI industry was worried about the wrong Chinese 'models' — 8gulmohar · 2026-11-09