Flux vs Vercel AI Gateway
Vercel AI Gateway puts models from many providers on one key at list price, with no markup, plus search, images, speech and transcription. You choose the model for every call. Flux picks a vetted model for each job and bills one flat rate per lane.
Checked 28 September 2026. See sources
Who each is for
Teams building on Vercel who want list prices, spend controls and logs, and are happy to choose each model in code.
People running coding agents and assistants who want the model chosen for them, from any tool that takes a base URL.
Side by side
| Capability | Flux | Vercel AI Gateway |
|---|---|---|
| Every major model on one key | Yes | Yes |
| Picks the right model for each job | Yes | No. You choose |
| Live web search and page reading | Partly. Page reading now | Yes |
| Voices, speaking and listening | Partly. Transcription now | Yes |
| Actions in apps | Partly. Coming to your key | No |
| Memory across every model | Partly. Shipping | No |
- Models
- Flux90+ models from 12 labs, on one key.
- Vercel AI GatewayModels from many providers, for text, images, video, speech, transcription, embeddings and reranking.
- How the model is chosen
- FluxSend flux-auto and Flux picks among vetted models for each kind of task, cost-aware. Or name any model in the catalogue yourself.
- Vercel AI GatewayYou name the model. Fallbacks and routing rules decide what serves a request when your first choice cannot.
- Pricing
- FluxOne flat rate per lane: Fast, Standard and Reasoning. flux-auto bills at the Standard rate, or the Fast rate when it sends your request to the Fast lane and a Fast model answers. Never more than Standard. A model you name bills at the rate on the pricing page.
- Vercel AI GatewayProvider list price with no markup or platform fee on tokens. Card payment fees may apply. A monthly free credit covers a subset of models.
- Cost of each call
- FluxIn the headers and body of every non-streamed response, with the model that answered.
- Vercel AI GatewayIn the request logs, and from an endpoint that returns the cost of one request.
Where each one wins
Where Vercel AI Gateway wins
- Zero markup on tokens, even with your own provider keys.
- Built into Vercel projects, where deployments can sign in without a key.
- Team controls: budgets per project, key or person, and model and provider allowlists.
Where Flux wins
- A model picked for each job. With Vercel you choose each model, or build your own router.
- One flat rate per lane, and flux-auto never costs more than the Standard rate, whichever model answers.
Which to choose
Choose Vercel AI Gateway if you build on Vercel, want list prices with no markup, and pick every model yourself.
Choose Flux if you want each job sent to a vetted model for you, at one flat rate per lane.
Questions
Is Vercel AI Gateway cheaper than Flux?
Per token, usually yes: it charges list price with no markup. Flux charges one flat rate per lane, and flux-auto bills at the Standard rate, or the Fast rate when it sends your request to the Fast lane and a Fast model answers. You pay Flux to choose the model for you.
Do I need to host on Vercel to use it?
No. Its API key works from any environment. Deployments on Vercel can also sign in without a key.
Can I use Flux from the AI SDK?
Yes. Flux speaks the OpenAI-style API, so the AI SDK works with a new base URL and key. The guide shows how.
Sources
- Vercel AI Gateway overviewChecked 28 September 2026
- Vercel AI Gateway pricingChecked 28 September 2026
- Vercel AI Gateway FAQChecked 28 September 2026
- Vercel AI Gateway web searchChecked 28 September 2026
- Vercel AI Gateway routing rules (2 July 2026)Checked 28 September 2026
We re-check these facts every quarter. If something here is out of date, tell us and we will fix it.
Try Flux with the agent you already use.
One key, any agent. Point the tool you already use at Flux and keep everything else.