16 language models, ranked on 309 real work tasks. Compare quality scores for document analysis, RAG search and chat, and orders and procurement, together with cost per 1,000 tasks and answer time. Sort the table, switch the chart between cost and speed, or drill down into individual task types. Results as of October 2026.
LLM Benchmarks
October 2026
Please note that this is the first version of our new LLM benchmark, so not every current model is included yet.
We are adding more models step by step, and future versions will cover all relevant current models.