Shared from Saga · AI signal
Operating modelRelevance · high2026-07-10
Vercel·4 min read

Enterprise AI usage data shows open‑weight models rising and prices stabilizing

TechnologyRetailFinancial ServicesHealthcare

Vercel’s July 2026 AI Gateway Production Index, based on June 2026 production traffic, reports that open‑weight models now account for 29% of token volume, while price per token has flattened even as total token volume and spend both grew around 27–29%. Anthropic models still dominate spend in high‑stakes use cases, but open‑weight adoption is accelerating for cost‑sensitive workloads.

Why it matters for leaders
For an AI/Data Lead shaping platform strategy, this is a signal that a mixed portfolio – combining premium closed models for high‑risk flows with cheaper open‑weight models for “good enough” tasks – is becoming the norm. Your architecture for MVPs should assume model swapping, not model lock‑in.
Opportunity signal

Design your first AI MVPs so that prompts, evaluation harnesses, and safety filters are model‑agnostic, allowing you to route specific workloads to lower‑cost open‑weight models as they mature and to premium providers for safety‑critical flows.

high
Risk signal

If your current stack hard‑codes a single model provider into application logic, you may be overpaying and unable to respond quickly if future pricing, throttling, or licensing terms change.

medium
Read the original source
The weekly signal

A visual weekly digest of the AI moves that matter

Curated signals, trends, and what they mean for leaders — one considered email a week. No noise.

Double opt-in. One-click unsubscribe in every email. We never share your address.

Where leaders make sense of AI

See the full picture inside Saga

This is one signal. Saga curates the AI developments that matter for your business, translates them into decisions, and takes you into a real-time strategy room — with a free, board-ready AI Leadership Report to start.