evaluation
6 articles
What the Agent Remembers
November 11, 2026 · 23 min read · agents, evaluation, cost
A note habit, a maintained wiki, standing instruction files: what persists for an agent to build on, and what upkeep earns its cost.
The Test Changed. What Did the Score Mean?
November 8, 2026 · 26 min read · evaluation, cost, security
A leaderboard moved without an update; a proof certifies a label, not an answer. What cited AI evaluation scores prove — and leave open.
October Deep Dive: AI Models — Promises & Breakthroughs
October 28, 2026 · 26 min read · generative AI, evaluation, cost
Thirteen AI model launches in 26 days. What October 2026 releases actually established — and the checklist for what you can verify a month on.
What Did DeepSeek Actually Open-Source?
October 27, 2026 · 15 min read · generative AI, cost, evaluation
DeepSeek open-sourced its DeepInfer engine, but withheld the caching and routing behind its revenue figures. Here is what the code confirms.
What Does OpenAI's Aurora Image-Model Testing Actually Establish?
October 27, 2026 · 10 min read · generative AI, evaluation
OpenAI's API briefly listed three Aurora image models in October 2026. What benchmark and community tests verify — and what remains unconfirmed.
Flint: a Laptop Coding Model Built by AI
October 25, 2026 · 11 min read · generative AI, evaluation
Recursive claims its Eureka Machine built Flint, an open-weight local coding model. What benchmark claims verify, and how close is AI-built AI?
