Skip to content
AI Credits
简体中文Sign upWorkspace
AI Credits

English

简体中文

English

简体中文
BUILD WITH AIExplore Chinese model integrations

Chinese & global AI.
One platform.
Ready to build.

DeepSeek, Kimi, MiniMax, Zhipu GLM, OpenAI, Claude, Gemini, and Grok. Connect through supported protocols and official SDKs, with usage tracked in Credits.

Explore plansRead the docs
Unified APIClear usage billingManual renewal
ONE CONNECTION. MORE POSSIBILITIES.
A shared workspace where ideas turn into applications
your-next-idea.ts
01 import OpenAI from 'openai'
02
03 const client = new OpenAI({
04 apiKey: process.env.AICREDITS_API_KEY,
05 baseURL: process.env.API_BASE_URL + '/v1'
06 })
07
08// Make room for your next idea.
AI Credits APIIntegration example
AI CreditsUNIFIED API
Text & chat
Multimodal
Tool calling

Capabilities depend on the published model catalog.

BUILT FOR YOUR DEVELOPMENT WORKFLOW
OpenAI-compatible
Anthropic-compatible
Gemini-compatible
SDK & Docs

MODELS, CONNECTED

Chinese and global models. One integration.

Chinese model integration guide →

DeepSeek, Kimi, MiniMax, and Zhipu GLM join OpenAI, Claude, Gemini, and Grok through platform credentials, compatible protocols, and Credits billing.

CHINESE MODELS

DeepSeek

Connect chat and reasoning models while preserving reasoning content and tool-call fields.

  • Chat
  • Messages
  • Responses
DeepSeek integration details →
CHINESE MODELS

Kimi

Connect Kimi / Moonshot models with model-specific context, thinking, and tool calling.

  • Chat
  • Messages
  • Responses
Kimi integration details →
CHINESE MODELS

MiniMax

Access text and reasoning capabilities using the request formats declared for each model.

  • Chat
  • Messages
  • Responses
MiniMax integration details →
GLOBAL MODELS

OpenAI

Use familiar conversation APIs and select embeddings or image capabilities from the catalog.

  • Chat
  • Responses
  • Messages
OpenAI integration details →
GLOBAL MODELS

Claude

Use Messages and Chat / Responses compatibility, with token counting where configured.

  • Messages
  • Chat
  • Responses
Claude integration details →
GLOBAL MODELS

Gemini

Use native generation, streaming, and token counting, or a declared compatible protocol.

  • Gemini
  • Chat
  • Messages
Gemini integration details →
GLOBAL MODELS

Grok

Connect conversation and reasoning, plus enabled image or video models from the catalog.

  • Chat
  • Responses
  • Messages
Grok integration details →

These are implemented integrations. Callable model IDs, protocols, streaming, and rates depend on the published catalog and your plan. A family name does not imply support for every API.

Service status

Service status is temporarily unavailable.

Data policy

Prompt and Response content is not stored by default; see the privacy policy for handling rules.

Privacy policy

Payment methods

Available methods, currencies, and taxes are confirmed at checkout.

View pricing

DESIGNED FOR DEVELOPERS

From model calls to a complete application experience.

Chinese and global models

DeepSeek, Kimi, MiniMax and GLM alongside OpenAI, Claude, Gemini and Grok. Check each published model’s protocols.

Conversation, retrieval, and creation

Build with reasoning, tools, embeddings, image or video generation, subject to model parameters and limits.

Official SDKs

TypeScript, Swift and Kotlin SDK 0.3.0 adds public model discovery. TypeScript supports SSE and cancellation.

Embedded user services

Use short-lived tokens for in-app balances, transactions, plan purchases, and support tickets.

Keys and access control

Configure model access, IP allowlists, and rate limits for API keys within your plan entitlements.

Credits and billing details

Review available balance, reserved Credits, and request usage alongside published rates and transaction records.

FROM ZERO TO YOUR FIRST CALL

Make your first call in three steps

Keep your familiar tools. Get started with the protocol your model supports.

  1. 01
    Register and purchaseChoose a plan and register; checkout retrieves the current price again.
  2. 02
    Create an API keyCreate and keep keys on a trusted backend; use short-lived tokens for device applications.
  3. 03
    Choose a protocol and modelConfirm an accessible model, configure the protocol URL, and send a request with an SDK or HTTP.
Read the quickstart →
curl https://api.example.invalid/v1/models \
  -H "Authorization: Bearer $API_KEY"
Keep your API key on your server.

Choose Chinese and global models for your application.

View all models and rates

MADE FOR WHAT YOU MAKE

Build something that is yours.

Start with your idea. Find the capabilities to match.

A developer workspace with a laptop and code on screen
APISDKAPPLICATIONS

01 / BUILD

Bring your application to life.

Use TypeScript, Swift, or Kotlin to connect model calls, user balances, purchases, and support in your application.

Explore the possibilities
People collaborating around laptops at a shared table
CHATTOOLSWORKFLOW

02 / ASSIST

Make conversations useful.

Build Chinese-language assistants and tool workflows with DeepSeek, Kimi, MiniMax or GLM; choose capabilities by model.

Explore the possibilities
A notebook and writing tools for planning creative work
VISIONIMAGESVIDEO

03 / CREATE

Give your ideas a head start.

Choose catalog-listed models for multimodal understanding, image generation, and video creation.

Explore the possibilities

Illustrative use cases. Available capabilities depend on the published model catalog.

How Credits work

Token usage uses the published unified rate; image, video, and per-request charges follow the units listed for each model. Costs and balances use Credits.

Reference face value of 1 Credit
0.001 USD
Example usage for 1M tokens
25 Credits
  • Credits expiring first are used first
  • Period plans do not auto-renew
  • Refunds apply only to an eligible unused portion
  • See the docs for failed-request billing rules
  • Expiry, rollover, and taxes follow the plan and checkout

PLANS THAT FIT YOUR NEXT STEP

Choose a plan for your current stage

View full pricing and rules
Published dataData as of:

Design preview · Sample plans and prices, not available for purchase.

Manual renewal · No automatic renewal · Final taxes shown at checkout

Period plans

monthly

Starter Preview

9 USD / monthly

tax added at checkout

1,000 Credits
  • 30 days validity
  • 1 eligible modelsfixture-text-standard
  • 60 RPM · 60,000 TPM
  • Concurrent requests 2
  • Top-ups supported · No rollover
monthly

Team Preview

29 USD / monthly

tax added at checkout

4,000 Credits
  • 30 days validity
  • 2 eligible modelsfixture-text-standard, fixture-multimodal-preview
  • 180 RPM · 180,000 TPM
  • Concurrent requests 6
  • Top-ups supported · No rollover

Top-up packs

Top-up packs

Top-up Preview

5 USD / one time

tax added at checkout

500 Credits
  • 90 days validity
  • 1 eligible modelsfixture-text-standard
  • 60 RPM · 60,000 TPM
  • Concurrent requests 2
  • Top-ups supported · No rollover

Service announcements

All announcements

GOOD TO KNOW

Frequently asked questions

Visit support center →
Which Chinese model families are integrated?

DeepSeek, Kimi, MiniMax and Zhipu GLM are integrated. DeepSeek, Kimi and MiniMax have Chat, Messages and Responses paths; GLM has Chat and Messages. Available models depend on the public catalog and your entitlements.

Do period plans renew automatically?

V1 period plans do not auto-renew. Checkout and the terms show current expiry and repurchase rules.

Can unused Credits be refunded?

Only an eligible unused portion may be refundable, subject to the refund policy and order state.

Are Prompt and Response contents stored?

Not by default. See the privacy policy and docs for security, logging, and exception handling.

Does protocol compatibility mean identical behavior?

No. Capabilities, parameters, and output behavior vary by model; review the compatibility guide.

Can every model use the same protocol?

Check protocol and streaming labels in the model catalog, then confirm accessible models with your credential. Model family names do not establish capability support.

How do I embed AI in my application?

Exchange short-lived user tokens on your trusted backend, then use the official SDKs for model calls, balances, transactions, and purchases.

YOUR NEXT CHAPTER

Your next idea starts with a call.

From an idea to your first API call. Build something that matters.

Get startedRead the quickstart