Claude Fable 5.1 joins the Ranking IA: first, by a hair
We ran Anthropic's new model through the 25 tests of the Ranking IA. It enters in first place with 97.3 %, tied with Opus 5 and at twice the price.
News
Data anonymization before AI: how micapa.ai does it
Pasting a contract into ChatGPT hands your client's data to someone else. Here is how we trained a model that masks it first, with real accuracy figures.
Spanish AI Benchmark: Eight Models Tested in Spanish
We tested eight frontier AI models in professional Spanish across 25 real office tasks. Who wins, where each one fails and why the overall score misleads.