OpenBMB's MiniCPM5-2B averaged 53.9 across its own benchmark set, ahead of larger 4B-class models. What the numbers mean, how Liquid AI's DSpark drafter speeds up small vision models, and why research on uncertainty features matters for deploying small models.
Anthropic released Claude Fable 5 on June 9, 2026, a public Mythos-class model whose safeguards route flagged cyber, biology and chemistry requests to Claude Opus 4.8, alongside Mythos 5 for vetted Project Glasswing partners. Pricing, vendor-reported benchmarks, the 30-day retention policy and what developers should check.
A practical guide to LLM evaluation: what the common benchmarks measure, why saturation, contamination and harness differences make leaderboard numbers unreliable, newer tests such as Humanity's Last Exam and METR time horizons, and how to build your own evaluation set.
A fact-checked comparison of Anthropic's Claude Opus 4.6 and OpenAI's GPT-5.2: context windows, published benchmark claims, effort and reasoning controls, agent features and API pricing, with guidance on which suits which workloads and a note on newer successors.