IO · 2026-11-05 10:04 UTC

Open-Weight Models Just Became the Default — Is Your Infrastructure Ready?

Open-weight models now process 71.5% of gateway tokens while accounting for 23% of spend, with spend share quadrupling since June 2026, according to Vercel's AI Gateway Production Index for October 2026. Western open-weight model releases from Meta, NVIDIA, Mistral, and others have removed provenance concerns that previously blocked enterprise adoption due to data-handling and jurisdiction constraints. Infrastructure teams must now treat open-weight inference as a core operational responsibility rather than a research experiment, managing serving infrastructure, version upgrades, and…

Loading the story on Tree News…