Skip to content
gentic.news — AI News Intelligence Platform
Connecting to the Living Graph…

Listen to today's AI briefing

Daily podcast — 5 min, AI-narrated summary of top stories

Bar chart titled Vercel AI Gateway Spend showing open model spend dropping sharply to 2.91% while Anthropic, OpenAI…

Vercel Data: Open Models Spend Collapses to All-Time Low

Closed AI models hit 97.09% spend share via Vercel; open models at all-time low over past 5 days.

·3h ago·2 min read··12 views·AI-Generated·Report error
Share:
What does Vercel AI Gateway data show about open vs closed model spending trends?

Open-source AI model spend through Vercel's AI Gateway fell to an all-time low in the past 5 days, while Anthropic, OpenAI, and Google combined hit 97.09% spend share on June 27, 2026, according to @rauchg.

TL;DR

Open model spend share hit all-time low · Closed models hit 97.09% spend share · Data from Vercel AI Gateway

On June 27, Anthropic, OpenAI, and Google captured 97.09% of AI model spend through Vercel's AI Gateway. Open-source model spending has since collapsed to an all-time low over the past five days, according to Vercel CEO Guillermo Rauch.

Key facts

  • 97.09% peak spend share for closed models on June 27
  • All-time low for open models in past 5 days
  • Data from Vercel AI Gateway spanning thousands of developers
  • Anthropic + OpenAI + Google combined share
  • Rauch: 'Open models spend was barely a thing'

The data, shared by Vercel CEO Guillermo Rauch in a tweet thread on July 2, 2026, tracks real-time inference spending across thousands of developers using the Vercel AI Gateway. The 97.09% share represents the combined spending on Anthropic's Claude, OpenAI's GPT-4o, and Google's Gemini models — the highest concentration of closed-model spend ever recorded through the platform.

"Open models spend was barely a thing," Rauch wrote, describing the June 27 peak. The subsequent five days saw open model spending hit an all-time low, though Rauch did not disclose exact percentage figures for the decline. Vercel's AI Gateway sits between developers and model providers, routing API calls and aggregating usage data across its customer base.

The trend contradicts the narrative that open-weight models like Meta's Llama 3 and Mistral are gaining enterprise traction. Vercel's developer-heavy user base appears to be voting with API calls for proprietary models, likely driven by reliability guarantees and ease of integration. The data does not capture on-premise or self-hosted deployments, which remain a stronghold for open models.

Why this matters

The timing is notable: Meta released Llama 4 in April 2026 and Mistral launched Mistral Large 2 in May, yet neither has dented closed-model spending share on Vercel's platform. The all-time low for open models suggests that even as new open-weight models hit benchmarks, actual production spend remains overwhelmingly proprietary.

Rauch did not specify which open models are included in the data or how Vercel classifies "open models." The company has not published a formal report or methodology for the dataset.

What to watch

Watch for Vercel to publish a formal AI Gateway spending report, expected in Q3 2026, which could reveal exact open-model share percentages. Also track whether Meta's Llama 4 enterprise adoption announcements shift the trend in subsequent months.

Sources cited in this article

  1. Vercel CEO Guillermo Rauch.
  2. Rauch
Source: gentic.news · · author= · citation.json

AI-assisted reporting. Generated by gentic.news from 2 verified sources, fact-checked against the Living Graph of 4,300+ entities. Edited by Ala SMITH.

Following this story?

Get a weekly digest with AI predictions, trends, and analysis — free.

AI Analysis

The Vercel data presents a stark reality check for the open-source AI community. Despite impressive benchmark scores and vocal advocacy, open-weight models are losing the battle for production inference spend — at least among Vercel's developer-heavy customer base. The 97.09% figure is remarkable not just for its magnitude but for its direction: it represents an all-time high for closed models, not a baseline. This data point is particularly interesting given the broader industry narrative. Meta has aggressively pushed Llama 4 as an enterprise-ready alternative, and Mistral has secured major European cloud deals. Yet Vercel's real-world usage data suggests these efforts have not translated into meaningful API spend share on one of the largest AI gateway platforms. The caveat is significant: Vercel's AI Gateway primarily serves web and frontend developers building AI features into consumer applications. This segment may naturally gravitate toward managed API services over self-hosting. Enterprise deployments of open models, particularly in regulated industries or on-premise settings, would not appear in this data. Rauch's framing — "open models spend was barely a thing" — is characteristically blunt and likely intended to provoke. The tweet thread lacks methodological detail, and Vercel has not disclosed whether the data is revenue-weighted or request-count-weighted, which could skew results if open models are cheaper per request. Still, the signal is clear: for the developer segment Vercel serves, proprietary models are winning on spend. The open-source community needs to either accept this reality or explain why Vercel's data is an outlier.
Compare side-by-side
Anthropic vs OpenAI
Enjoyed this article?
Share:

AI Toolslive

Five one-click lenses on this article. Cached for 24h.

Pick a tool above to generate an instant lens on this article.

Related Articles

From the lab

The framework underneath this story

Every article on this site sits on top of one engine and one framework — both built by the lab.

More in Opinion & Analysis

View all