Anthropic's per-token cost runs 4.4 times the average on Vercel, and developers keep paying

August 18, 2026 | Source: THE DECODER

Tags: Anthropic, Vercel, Claude Fable 5, AI Gateway, pricing, LLM costs

Vercel's July AI Gateway data reveals Anthropic captured 65.1% of spending while handling only 30% of tokens — its per-token cost runs 4.4x the platform average. Claude Fable 5 claims 13.2% of spend with 9-in-10 teams being new customers, signaling premium models retain pricing power even as average token prices fell 13.6% industry-wide.

Details

Anthropic dominated Vercel's AI Gateway in July, pulling 65.1% of total revenue from just 30% of tokens processed. Its per-token cost runs 4.4 times the average of all other providers on the platform — a premium that developers keep paying. Claude Opus 4.8 is the top spending model; Fable 5 came in second at 13.2% of Gateway spend. Fable 5's growth dynamics are notable: nine in ten Fable 5 teams in July were new customers, which Vercel interprets as fresh demand rather than internal model upgrades. This counters Ramp's earlier read that Fable might be overpriced — from Vercel's vantage, developers are actively choosing it over alternatives. The broader market is moving in two directions simultaneously. Token volume rose 59% in July while spending grew only 37%, meaning average prices fell 13.6% as usage shifted to cheaper tiers. For the first time, xAI and Moonshot captured a visible share of open-weight model spend on Vercel — a sign the market is diversifying beyond the top three providers. Both Vercel and Ramp see only slices of the full AI deployment picture. Vercel's data reflects developer-facing hosted APIs; enterprise direct contracts and on-premise deployments remain invisible. Still, the 4.4x cost premium Anthropic commands on this platform is a concrete signal that Claude's performance advantage is currently justifying its pricing in the developer market.