Vectrust
Get an API key

Every model,one API.

Models from OpenAI, Anthropic, Google and the leading Chinese vendors, on one account and one bill. Native formats: your code changes only its base URL.

New on Nov 2Muse video and podcast, TypeSafe jev, four chat models
  • Live within 30 min of release
  • Native OpenAI · Anthropic · Gemini
  • Public status page
500T
Tokens a month
180B+
Requests served
200+
Vendors
800+
Models
Get an API key
api.openai-next.com
Same request, any modelExample
POST/v1/chat/completionsmodelclaude-opus-5-5
Type a message…
Send

Vendor by vendor vs. OpenAI Next

Vendor by vendor

Sign up and verify with each vendorOnce per vendor
Top up and track a balance at eachOne per vendor
Adapt to each SDK and request formatOne per vendor
Apply for quota with eachPer vendor
Adapt and retest for new modelsEvery release
Reconcile each vendor's invoiceMonthly, per vendor
With N vendorsEvery line above × N

More vendors, more upkeep.
And every new model starts the cycle again.

OpenAI Next

1

API key

One account
One balance
Native formats
Live in 30 min
Get an API key

One endpoint for the leading models

Leading vendors from China and abroad. Text, image, video and audio, all called the same way.

Four modalities

Chat, images, video, music and speech, all on one API key.

Text
Claude Opus 5.5, DeepSeek V4 Pro
Image
GPT Image 2.5, Nano Banana Pro
Video
Seedance 2.5, Kling 4.0
Audio
Suno v6, TTS-1 HD
Parameters and samples for every model in the API reference →

Leading vendors

OpenAI, Anthropic, Google and xAI, plus Chinese vendors such as DeepSeek, Qwen, Zhipu, Kimi, Doubao and MiniMax.

OpenAI
Anthropic
Google
xAI
Meta
DeepSeek
Qwen
Zhipu
Kimi
Doubao
MiniMax
Kling
ERNIE
Hunyuan
Suno
The full list is on the models page →

Native formats

Use the OpenAI, Anthropic and Gemini SDKs as they are. Only the base URL and API key change.

Live in 30 minutes

New models go live within 30 minutes of the vendor's release, on the API key you already have.

Spend controls

Each API key has a group, a multiplier, a spending cap and an expiry. Every charge is itemized in the console.

Async tasks

Slow jobs like video and music are submitted, then polled for the result. Every accepted task has a status you can check.

Get an API key

What you can build

Try each one on NextLabs, parameters and code included.

Text

Customer support and knowledge Q&A

Connect your knowledge base and answer customers and staff from your own documents, handing off to a person when needed.

Suggested: DeepSeek V4 ProTry it →
Text

Code generation and review

Write functions, add tests, review pull requests. Tools like Claude Code and Codex connect directly.

Suggested: Claude Opus 5.5Try it →
Text

Contract and long-document review

Read a whole contract or tender in one pass, flag risky clauses and list the inconsistencies.

Suggested: Gemini 3.1 ProTry it →
Text

Data analysis and reporting

Hand over spreadsheets and business data; get findings, chart notes and a first draft of the weekly report.

Suggested: GPT-6.1 SolTry it →
Text

Copy and marketing content

Product descriptions, campaign copy and social posts in your brand voice, several versions at a time.

Suggested: Kimi K3Try it →
Text

Translation

Translate documents, UI strings and support chats, keeping the formatting and your glossary.

Suggested: Qwen3.5-PlusTry it →
Image

Posters

Posters and key visuals from a single sentence, sized for each channel.

Suggested: GPT Image 2.5Try it →
Image

Product photos

White-background, lifestyle and on-model shots of one product, in batches to each channel's spec.

Suggested: Seedream 5.0 ProTry it →
Image

Storyboards

Shot by shot from a script, with reference images to keep characters consistent.

Suggested: Nano Banana ProTry it →
Video

Promo videos

Short product and brand videos from a paragraph or a first frame.

Suggested: Seedance 2.5Try it →
Audio

Music

Background music and songs with vocals, by style, mood and length.

Suggested: Suno v6Try it →
Audio

Voice-over

Turn a script into natural speech for narration, courses and announcements.

Suggested: TTS-1 HDTry it →
More use cases and recipes ↗

Three ways to connect, compared

Model coverage

Direct to each vendorThat vendor's models only
Your own gatewayWhichever upstreams you connect
OpenAI NextLeading vendors from China and abroad; text, image, video, audio

Integration

Direct to each vendorA different SDK and format per vendor
Your own gatewayUsually one unified format
OpenAI NextNative formats: OpenAI, Anthropic and Gemini all work

New models

Direct to each vendorAdapt to each vendor's releases
Your own gatewayYou track and test them yourself
OpenAI NextLive within 30 minutes of release

Accounts and billing

Direct to each vendorSign up, verify and top up at each
Your own gatewayStill an account at every upstream
OpenAI NextOne account, one balance

Usage controls

Direct to each vendorA separate console per vendor
Your own gatewayBuild it yourself
OpenAI NextGroups, multipliers, spending caps and expiry per API key

Long requests

Direct to each vendorTimeouts differ by vendor
Your own gatewayConfigure timeouts and retries yourself
OpenAI NextNo 100-second limit on the US node; video and music run as async tasks

Operations

Direct to each vendorNothing to deploy, but many status pages to watch
Your own gatewayDeploy, scale and stay on call yourself
OpenAI NextNothing to deploy; a public status page

Docs

Direct to each vendorDifferent docs for every vendor
Your own gatewayWrite your own
OpenAI NextParameters, samples and error codes for every model on NextLabs
Get an API key

For procurement and long-term agreements, contact sales.

Your tools, one URL away.

Claude Code: set ANTHROPIC_BASE_URL to https://api.openai-next.com

Codex: add a provider in config.toml that uses the Responses API

Cursor: override the OpenAI base URL in the model settings

Cherry Studio, LobeHub: add an OpenAI-compatible provider

Dify, FastGPT: enter the base URL and API key under model providers

Gemini CLI: point GOOGLE_GEMINI_BASE_URL at the platform and keep the native Gemini format

Read the setup guide

Long requests, new models, outages: how each is handled

100 s

Long requests

The main address sits behind Cloudflare, which cuts any request off at 100 seconds with a 524. Image and video generation hit this most.

Switch to the US node us.api.openai-next.com: a direct connection without that limit, with the same routes and API key.

30 min

New models

When a vendor ships a new model, a direct integration has to enable, adapt and test it all over again.

It goes live on the platform within 30 minutes of the official release, on the API key you already have.

24 h

Live run data

When something fails, it's hard to tell whether it's your code, the network or the upstream.

Incidents are posted on the public status page, and the NextLabs status page tracks each model's success rate and latency over the last 24 hours of real runs.

View service status ↗
View service status

Choose a console by how you pay

Each console has its own sign-in, and API keys don't carry across.

Default

Usage-based

api.openai-next.com

Per token

Chat priced per million tokens; some media models per call

Get an API key
  • All four modalities: text, image, video, audio
  • Input and output priced separately; every charge itemized in the console
  • Groups, multipliers, spending caps and expiry per API key
  • US node us.api.openai-next.com, so long requests aren't cut off
  • Native OpenAI, Anthropic and Gemini formats

Per-call

draw.openai-next.com

Per call

One image, one video or one song is one call

Open the per-call console
  • Fixed-price work: images, video, music
  • Same request formats as the usage-based console
  • Separate accounts and API keys; not shared between the two

Enterprise

yun@opendevelop.tech

Custom

Procurement, contracts, long-term agreements

Contact sales
  • Enterprise and institutional procurement
  • Partnerships with model vendors
  • Integration planning

Recently added

New models go live within 30 minutes of their official release. Newest first.

Meta Muse: text and image to video, first/last frame, 20–60 s extensions, multi-voice podcasts; verified on both platforms. 16:9 and 9:16 only.

Muse Video7 models

Meta · Nov 2

Try it ↗API reference ↗

TypeSafe jev: a decision model with structured output, no free text; output tokens are free. Best in English.

T

Jev2 models

TypeSafe · Nov 2

Try it ↗API reference ↗

Zhipu GLM-5.3: global is the full-parameter overseas build, flashx the fast one.

glm-5.3-global2 models

Zhipu AI · Nov 2

Try it ↗API reference ↗

Moonshot Kimi K3.1 Pro: 1M context, built for long-horizon agents and tool calls.

Kimi K3.1 Pro

Moonshot AI · Nov 2

Try it ↗API reference ↗

OpenAI GPT-6.1 Terra, on the api platform.

GPT-6.1 Terra

OpenAI · Nov 2

Try it ↗API reference ↗
See all models

Frequently asked questions

Chat models bill per token, with input and output priced separately per million tokens. Image, video and music models bill per call or per token. Each model's price is on its NextLabs model page. A task the platform has accepted is billed whether or not you collect the result.

api.openai-next.com (usage-based) is the default and carries models in all four modalities. draw.openai-next.com (per-call) is a separate site with the same request formats, but its own accounts and API keys: a key from one site gets a 401 on the other.

The native OpenAI Chat, OpenAI Responses, Anthropic Messages, Gemini, OpenAI Images and OpenAI Speech formats; only the base URL and API key differ. Video and music models work as async tasks: submit, poll, collect the result. See Base URL & auth.

Within 30 minutes of the vendor's official release, on the API key you already have. Recently added on this page lists the latest.

You create API keys on the console's token page, as many as you need. Each key gets a group when you create it, and the group sets its price multiplier and which models it can call. A key can also have a spending cap and an expiry.

The main address sits behind Cloudflare, which cuts off any request that runs past 100 seconds. Switch to the US node us.api.openai-next.com; the routes and API key stay the same.

For procurement, contracts and long-term agreements, contact sales directly. See Enterprise.
Get an API key
Vectrust

Start with one API key

Sign up, add credit and create an API key to call every model on the platform.

Get an API keyEnterprise