Vercel index: open-weight models carry 71.5% of gateway tokens, 23% of spend

Vercel's October AI Gateway index says open-weight models now handle over seven in ten tokens and nearly a quarter of spend, and team switching hit a record.

Vercel's AI Gateway Production Index for October 2026, published October 15, reports that open-weight models now process 71.5% of gateway tokens and take 23% of spend. Open weight's share of spend has more than quadrupled since June. For builders, the more useful number is churn: in September almost one in three teams changed their most-used lab, an index record, and nearly one in two changed their most-used model.

Vercel says the cheapest models drive token volume and the expensive ones drive spend. DeepSeek-v4.1-Flash alone processed nearly half of September's tokens for 2.6% of spend. Overall, September spend was about 13 times December's while average price per token fell 47%. Frontier spend grew tenfold over that period even as its share shrank. Anthropic's share of gateway spend fell from 63.7% to 43.9%. Teams processing at least 10 million tokens used an average of 15 models since December, rising to 42 for billion-token teams.

The limits matter. Vercel estimates spend at published list prices, so actual bills may differ. TechCrunch notes the gateway is a limited view that may undercount labs like OpenAI and Anthropic, which sell through their own APIs. The report doesn't say how many teams are in the sample.

A second gateway points the same way on volume. DeepSeek V4.1 Flash tops OpenRouter's rankings by tokens, though OpenRouter says those rankings measure neither users nor spend. Digital Applied warns against reading gateway shares as a market census. It suggests treating adoption as a reason to test a model on your own tasks, with a fallback in place.

In practice: if your code assumes one provider, the data suggests your peers are already swapping models. A routing layer and a small eval set make those swaps cheap.

Sources