#benchmarks

GPT-6.1 Terra: OpenAI’s cheaper model for doing work on a computer

Roman Aglaganov29 OCT 2026

Mercury 3 writes the whole answer at once, and fast

Roman Aglaganov22 OCT 2026

Recursive’s Flint: a free coding model for laptops, designed largely by AI

Roman Aglaganov21 OCT 2026

Gemini 4 Argon opens up: Google’s frontier model is now on sale to everyone who pays

Roman Aglaganov14 OCT 2026

Muse Wonder: Meta’s first frontier model wants to remember you

Roman Aglaganov14 OCT 2026

Microsoft-Decision-1: a model that answers with odds, not essays

Roman Aglaganov14 OCT 2026

Claude Haiku 5.5: a small model that got cheap and sharp

Roman Aglaganov10 OCT 2026

Gemini 1.5 and the million-token memory

Roman Aglaganov09 OCT 2026

Opinion: AI benchmarks are broken, and we keep using them anyway

Roman Aglaganov09 OCT 2026

GPT-4: the model that sat the bar exam

Roman Aglaganov09 OCT 2026

Opinion: Every AI should learn to say “I don’t know”

Roman Aglaganov09 OCT 2026