Anthropic Beats OpenAI 2-to-1 With New Teams | Harpreet Arora artwork

Anthropic Beats OpenAI 2-to-1 With New Teams | Harpreet Arora

MTS

November 13, 2026

Vercel AI Product Lead Harpreet Arora shares enterprise adoption trends from Vercel’s AI Gateway, revealing how teams are shifting toward Anthropic models, decision models, and voice workflows.

0:00

Harpreet Arora

With Anthropic, what we're seeing is, in the last month, new teams that were choosing the AI Gateway were using Anthropic models at twice the rate of OpenAI, for instance. And so if you go all the way back to the beginning of the year, Opus 4.5 came out from Anthropic, and it just all hell broke loose. And then they just had this dominating streak till April of this year or May, and then you started seeing OpenAI models come back and take share. And then Anthropic kind of had a lull in August and September. But Opus 5.5, such a great model.

And I think they've doubled down with Fable 5.2. And you're seeing that newer teams are just experimenting with Anthropic models quite a bit.

0:45

SPEAKER_2

Hello, everyone. Welcome back to MTS. Today, we are joined by Harpreet Arora, an AI product lead at Vercel. So Harpreet, welcome again to MTS.

0:55

Harpreet Arora

Thank you for having me. It's good to be here this time. I'm directly in your office.

0:59

SPEAKER_2

Directly in the office? Instead of Vercel ship, we're changing spots.

1:02

Harpreet Arora

We get all this background noise, but it's just still here. It's nice.

1:05

SPEAKER_2

Yeah, I know. Last time we talked, really interesting stuff because you lead a lot of the AI products at Vercel, specifically when it comes to AI Gateway. It gives you a lot of insights about what's happening at the frontier, how companies, enterprises, customers are using different models. What trends are you noticing?

1:24

Harpreet Arora

Yeah, for sure. I mean, just to educate everyone, Vercel Gateway sits at this unique point where we work with the labs, we work with providers of open-weight models, we work with the open-weight labs, we work with customers. So we have this unique insight into how people are using AI.

One of the main trends, and this is something we talked about in our report as well, is it's hard to keep up with the pace of change. Decision models, for instance, were not a thing on anyone's radar in January of this year.

And now it's like doing a quarter of our requests, right? And you could be like, oh, this is like a quarter of the requests. But what we're seeing is people are rethinking their workflows. So they are evaluating their pipeline and asking themselves, does everything need to go through a regular LLM? Is there something that can actually be made into a decision? Are there parts of my app that could use a voice agent, for instance, right? And so people are just re-innovating, and anyone who is taking the innovator's dilemma in stride is riding the wave. And so that's one thing. We're seeing trends with Anthropic growing. We're seeing trends with open models still continuing, but not at the pace that we saw. So happy to go into either of those.

2:44

SPEAKER_2

Any numbers you can share?

2:45

Harpreet Arora

So with Anthropic, what we're seeing is in the last month, new teams that were choosing the AI Gateway were using Anthropic models at twice the rate of OpenAI, for instance. And so if you look at, if you go all the way back to the beginning of the year, Opus 4.5 came out from Anthropic and it just all hell broke loose, and then they just had this dominating streak till April of this year or May, and then you started seeing OpenAI models come back and take share, and then Anthropic kind of had a lull in August and September, but Opus 5.5, such a great model, and I think they've doubled down with Fable 5.2, and you're seeing that newer teams are just experimenting with Anthropic models quite a bit.

3:32

SPEAKER_2

Yeah. I think the news from Jev was really interesting, seeing how AI Gateway, a lot of these enterprises all of a sudden wanted to use decision models for how they were interacting in using AI as different parts of their flow. You're also seeing this, of course, with OpenWeights model. Traditionally, the US has kind of been behind China when it came to being at the frontier of OpenWeights model, and the hope was with Nvidia and all of the work they were doing, we could be a leader. What's your take now?

4:06

Harpreet Arora

I mean, today is obviously a very interesting day to talk about it.

There's a lot that we don't know about that happened at Nvidia, so obviously we can't speculate there, but it is a setback, right? Jensen was a major voice for the OpenWeight models in the US as well as internationally, and we have seen, for instance, OpenWeight models were gaining a lot of share, and they're still in use with enterprises, but we haven't seen the same level of acceleration and model quality the way we were seeing before. Not to mention, most of the frontier OpenWeight models are coming out of China.

Read the other 11 minutes of this episode here.

$15 gets you this episode and 99 more. Credits never expire. After checkout, one link brings you back here to read it.

Not what you expected? Full refund within 14 days.

Thousands of transcripts fetched by people building searchable podcast archives

Fetch transcripts with code

The same credits work from a script or an AI agent. The demo key returns a sample episode in full, no card needed:

request
curl -H "x-api-key: pt_demo" \
  https://spoken.md/transcripts/1000651996090

Markdown with the speakers named, for your notes, your knowledge base, or anything that makes HTTP calls.

From $0.10 per transcript. No subscription. Credits never expire. Prices exclude VAT, added at checkout for EU customers. Not what you expected? Email us within 14 days with 20 or fewer credits used and we refund the pack in full.

Using your own key:

request
curl -H "x-api-key: YOUR_KEY" \
  https://spoken.md/transcripts/1000801656115