Vercel AI Gateway vs OpenRouter: which is better?
Vercel AI Gateway is the cheaper way to buy tokens at list price, with no platform fee and a polished Claude Code setup, while OpenRouter has the bigger free tier, cheaper own-key use and an automatic router. For coding agents there is a third option: Standard Compute, a flat plan with included compute that stops at its budget and a router that picks an efficient model per request.
Vercel AI Gateway and OpenRouter side by side
| Vercel AI Gateway | OpenRouter | |
|---|---|---|
| Fee | None on tokens; card processing fees on credits | 5.5% on card credit purchases ($0.80 minimum), 8% on Business |
| Your own keys | No fee, paid tier | Free up to $25,000 of list-price use a month, then 5% |
| Free tier | Monthly free credit on a subset of models; ends once you buy credits | 25+ free models, 50 requests a day, 1,000 after buying $10 of credit |
| Credits | Expire one year after purchase | May expire one year after purchase |
| Model choice | Provider order by cost, speed or throughput; rules you write | openrouter/auto picks a model per prompt; pareto-code picks the cheapest above a coding score |
| Spend caps | Per team, project, key or member; a soft cap | Credit limit per key with resets; workspace budgets on Enterprise |
| EU | EU or US inference per request at the provider's regional rate | EU and US in-region routing on Business and Enterprise |
| Claude Code | Dedicated endpoint, one-command setup, Claude subscription passthrough | Native, base URL openrouter.ai/api; guaranteed only with Anthropic as provider |
From vercel.com/docs/ai-gateway, openrouter.ai/pricing and openrouter.ai/docs, checked October 13, 2026.
What a coding-agent month costs
Pinned to one Claude model, Vercel is about 5% cheaper than OpenRouter, both beat a Standard Compute plan in a typical month, and the plan wins in a busy one. OpenRouter's advantage is that it can choose the model: its auto router ranks models by what the community spends on each task type over the past seven days, while Vercel's guide says the classifier is yours to build.
| Month of coding-agent use | Vercel AI Gateway | OpenRouter | Standard Compute |
|---|---|---|---|
| Typical days, Claude Sonnet 5.5 | $71 | $74 | Standard, $89 ($95 budget) |
| Typical days, Claude Opus 5.5 | $141 | $149 | Pro, $249 ($269 budget) |
| Busy days, Claude Sonnet 5.5 | $579 | $611 | Pro Plus, $499 ($549 budget) |
Month = 20 working days of the median (typical) or 90th-percentile (busy) coding-agent day in our Agent Burn Index. Vercel before card fees; OpenRouter with its 5.5% fee. Both list Sonnet 5.5 at $2 and $10 per million tokens and Opus 5.5 at $4 and $20.
Which to choose
- Choose Vercel AI Gateway for list-price tokens with no platform fee, apps already on Vercel, or Claude Code on your Claude subscription with Vercel's logs.
- Choose OpenRouter for the widest catalog, a free tier for experiments, an automatic router, or your own provider keys up to $25,000 a month at no fee.
- Choose Standard Compute for coding agents on a fixed monthly price: compute included, routing for cost, a plan per key for teams and agencies, and requests that stop at the budget.
The catch
Neither caps a month on its own terms: both bill per token, Vercel's budgets are a soft cap, and OpenRouter's key limits are yours to set. Put a limit on every agent key before the first long session.
Where Standard Compute fits
One key for Claude Code on /v1/messages and Codex on /v1/responses, with a router that chooses an efficient model for routine steps and keeps frontier models for hard ones. Plans start at $19 with the compute included, stop at their budget with no overage, and come as a plan per key for teams and agencies, with Europe or US rules on any plan.
FAQ
Is Vercel AI Gateway cheaper than OpenRouter?
For tokens bought on a card, yes, by about 5%: Vercel adds no platform fee and OpenRouter charges 5.5% on credit purchases. For your own provider keys, OpenRouter is free up to $25,000 a month and Vercel charges no fee either.
Does Vercel AI Gateway have an auto router like OpenRouter?
No. Vercel orders providers for the model you choose and applies rules you write; its routing guide says "the classifier is the part you own". OpenRouter's openrouter/auto picks a model per prompt.
Which is better for Claude Code?
Vercel has the smoother setup: a dedicated endpoint, a one-command installer, model discovery and Claude subscription passthrough. OpenRouter's cookbook works natively too but is only guaranteed with Anthropic as the provider.
Which has better EU support?
Vercel lets any request ask for EU or US inference at the provider's regional rate. OpenRouter's in-region routing needs a Business or Enterprise plan.