Vercel reported on 15 October 2026, in its AI Gateway Production Index for October, that TypeSafe AI’s Jev was the fastest-adopted model in AI Gateway history. In its first two weeks on the gateway (launch 17 September), Jev reached 35% of paid teams — 35.3% in Vercel’s launch table.
These numbers are Vercel AI Gateway metrics only. They describe traffic routed through Vercel’s gateway, not the global inference market, and they under-count labs’ direct APIs. This is a new post about gateway adoption, not an update to TypeSafe’s earlier funding story.
Adoption versus spend
Vercel’s scorecard puts Jev first on reach among September launches with at least two weeks of data — nearly double Claude Sonnet 5.5 in second place on that metric. Of teams that tried Jev in week one, 55.6% returned in week two (fifth-highest retention in the September set).
Spend tells a different story. Over that same first-two-weeks window, Jev accounted for just 0.27% of gateway spend. For September as a whole, Vercel says Jev was about 1% of the month’s tokens and 0.15% of spend. Vercel explains that Jev is a decision model: requests are faster, cheaper and use a fraction of the tokens of typical language-model calls, so broad trial does not equal a large inference bill.
How Vercel measures launches
Launch metrics cover external paid teams during each model’s first 14 days. Reach uses paid teams active from the start of the launch month through the data cutoff. Spend share uses total paid gateway spending during the model’s window. Return rate is the share of first-week teams that used the model again in week two. Spending in the index is estimated at published list prices.
Against that yardstick, Jev topped September reach; Claude Opus 5.5 led spend share in its own first two weeks (11.03% of gateway dollars), which Vercel calls the largest spend share of any September model launch.
Wider October index colour (context only)
The same report says open-weight models processed more than seven in ten gateway tokens and nearly one in four dollars in September, and that September gateway spend was about 13 times December’s while average price per token fell 47%. That market backdrop helps explain why a cheap decision model can win trial share without moving spend share — but the Jev claims that matter for this post remain the gateway adoption and spend figures above.
Not a global “most used model” claim
Do not read “fastest-adopted in AI Gateway history” as “most used AI model worldwide.” Vercel is clear the index is anonymised aggregate traffic through AI Gateway through September 2026. Teams that call TypeSafe (or anyone else) directly never appear.
Related TSN coverage
TSN previously covered TypeSafe AI’s Jev $870m Series A and explained RAG vs Jev-RAG. For another specialised decision-style model in the news, see IBM Granite Switch 4.1. Those are separate stories; today’s item is gateway telemetry, not a funding update.
What’s still unknown
- How many Jev tokens ran outside Vercel AI Gateway in the same window
- Whether TypeSafe agrees with Vercel’s characterisation of Jev as a decision model for billing purposes
- Retention beyond week two
What this means
- Vercel: Jev was the fastest-adopted model in AI Gateway history — ~35% of paid teams in two weeks.
- Same window: only 0.27% of gateway spend; September: ~1% tokens / 0.15% spend (Vercel).
- Scope: gateway traffic only.
Sources
Related stories
- TypeSafe AI Raises $870m Series A for Jev
- RAG vs Jev-RAG Explained
- IBM Granite Switch 4.1 Decision Speech Model
Follow @tsnmediahub on X for breaking AI news as it happens.
Follow on X
