The Cline Provider Now Runs on the Vercel AI Gateway

When you select the Cline provider in your IDE, your requests now route through Vercel AI Gateway. This change brings faster response times, better global coverage, and zero markup fees while maintaining the same seamless experience you've come to expect.

Built for speed and savings

The move delivers immediate, measurable improvements. From a 50/50 traffic split between our previous infrastructure and Vercel's gateway, production testing shows P99 streaming latencies improving 10-14% across our most popular models. Grok-code-fast-1 saw P99 streaming nearly 40% faster; Minimax M2 showed over 40% faster P99 streaming. Error rates are down 24.3% (from 1.81% to 1.37%).

Editorial note (updated): An earlier version of this post reported a 43.8% reduction in error rates (from 1.78% to 1%). That figure was based on a smaller production sample. We’ve since updated this metric to reflect a two-week measurement window during a 50/50 traffic split, which more accurately represents steady-state performance.

Vercel's global network spans more than a hundred points of presence worldwide. Your requests hit the nearest location, then travel across Vercel's private backbone to the AI Gateway. This architecture adds sub-20ms of routing overhead to your model requests.

For a technical breakdown on the Vercel AI Gateway, refer to their blog.

The economic benefits are equally straightforward. Vercel charges 0% markup on inference costs. When you use the Cline provider, you pay exactly what the model provider charges, nothing more. This aligns with our core philosophy: transparent pricing that puts developers first.

Shared open source values

This partnership makes sense beyond the technical improvements. Both Cline and Vercel built our businesses on open source foundations. We share a belief that infrastructure should be transparent, performant, and accessible to everyone, not locked behind opaque platforms that profit from hidden margins.

Vercel's decision to offer inference at cost mirrors our own approach. We don't believe in markup-based business models that tax every API call. Instead, we focus on building tools that developers want to use, not ones they're forced to pay premiums for.

What changes for you

Nothing about how you use Cline changes. The Cline provider works exactly as before, same models, same interface, same reliability. You might notice faster response times and lower costs, but the experience remains seamless.

All existing functionality carries over. Your API keys, model preferences, and project settings stay intact. The transition happens automatically; no action required on your end.

Cline supports over 40 API providers and local models. If you prefer Anthropic directly, OpenRouter, Ollama, or any other provider, those options remain fully available. The Cline provider is one choice among many.

Launch promotion: Free GLM-4.6

To celebrate this move, we're offering free access to GLM-4.6 through the Cline provider for a limited time. This model excels at complex reasoning tasks and multi-step coding operations. Try it now at no cost and experience the improved performance firsthand.

If you already have a Cline account, go ahead and start using GLM-4.6 for free. If you don't, create a Cline account here, and select glm-4.6 from the Cline provider.

The Cline provider will continue evolving with new models and capabilities. Our partnership with Vercel ensures we can deliver these improvements at scale while maintaining the open, accessible approach that defines Cline.