Anthropic is taking a disproportionate share of spending on Vercel's AI Gateway, according to The Decoder, which reports that the company accounted for 65.1% of Gateway spending in July while processing 30% of tokens. The same report says Anthropic's per-token cost on the platform was 4.4 times the average of other providers. That framing matters because the data is platform-specific. The figures describe usage and spending inside Vercel's AI Gateway, not the full AI model market. But for developers building through Vercel's routing layer, the July numbers suggest many teams are still willing to pay more for premium models when they decide the output quality or workflow fit justifies the cost. The Decoder also reports that Fable 5 quickly reached 13.2% of Gateway spending, putting it behind Opus 4.8 in that dataset. Nine out of ten Fable teams in July were new customers, a signal Vercel reads as evidence that the model is bringing in fresh users rather than merely shifting existing usage across providers. There is a counterpoint in the same report. Payments company Ramp had previously cited relatively low Fable usage as a sign the model might be priced too high. The Decoder notes that both Ramp and Vercel see only a slice of the broader market, so the two snapshots should not be treated as a universal verdict on demand. The wider Vercel Gateway data points to a market moving in two directions at once. The Decoder reports that token volume rose 59% overall and spending grew 37%, while average token prices fell 13.6% as companies moved toward cheaper tiers. In other words, usage is expanding, but buyers are also optimizing cost where they can. The useful read for operators is not simply that one provider is expensive. It is that model choice is becoming a budget line with visible trade-offs: premium models can command a large share of spend even when they represent a smaller share of token throughput, while cheaper tiers can still pull down average prices as usage scales. Because this cluster has only one source, the numbers should be treated as a credible Vercel-specific snapshot rather than a fully corroborated market benchmark. The main takeaway is narrow but meaningful: within Vercel's AI Gateway in July, Anthropic monetized usage at a much higher rate than competing providers on average, and developers continued to spend heavily on it. Who benefits: Premium model providers benefit if developers keep paying higher per-token prices for perceived performance. Vercel also benefits from having Gateway data that can help teams compare model economics inside its platform. Who's exposed: Teams standardizing on higher-cost models are more exposed to AI budget growth as usage scales. Competing providers are exposed if lower prices are not enough to displace premium models in high-value workflows.