An Open Field Guide to Inference Engineering · Modelplane Docs

Platform engineering teams are increasingly expected to run open models in-house as coding assistants, rising frontier-model bills and data-residency needs make hosted APIs impractical, according to a Modelplane documentation guide dated October 2026. Hugging Face now hosts more than two million open models, and DeepSeek's R1 matched OpenAI's o1 on math and coding benchmarks roughly a month after o1's release, the guide notes. Upbound, the company behind Crossplane and Modelplane, documented its own self-hosting effort in its Project Champagne blog series, running models on Modelplane. The guide matters because it frames inference engineering as the natural next role for platform teams, and points to watch include how quickly those teams adopt open-model serving stacks and whether the 10x annual cost decline continues.