LLM Scorebook is a technical publication for engineers and people learning to build with AI. We connect model developments to the decisions that follow: which configurations to evaluate, what a workload may cost, and how to implement useful workflows.
What we publish
Our library includes source-linked release analysis, model references, cost explanations and practical guides with inspectable examples. The leaderboard is a supporting reference, with a dated selection and visible measurement conditions.
How we handle evidence
We distinguish official claims, independent evaluations, calculations and our own executed tests. Missing results remain unknown. Some first-release evidence comes from the supplied research snapshot; the articles disclose its date and their verification scope.
Read and get in touch
Explore the journal, guides and methodology. Contact hello@llmscorebook.com with questions, corrections or a proposal.