does anyone have a benchmark for BAML vs openai function calling on gpt-5.1-mini/ gpt-5.1-nano or recent models for accuracy for classification . i need to convince the dependency in my company

hi @Sandipan let me share some data with you, but here's a few quick questions that may help me: 1. what language are you calling openai from? 2. what kind of classification are you doing (<50 categories, 100s of categories, 1000+ categories, 10000+ categories) each of these is completely different in terms of how to best approach it 3. are you able to install baml today on your work machine? usually the best benchmarks are those that are side-by-side comparisons of existing results -> those boosted by BAML, so if you're able to share, i'd love to get you something that would help you build the most compelling case. if its easier, you're also welcome to email me and we can get on a call or just dm me. (vbv@boundaryml.com)