Standard-compatible protocols. Credits never expire. Online top-up and key delivery. Access DeepSeek, Qwen, Kimi, GPT, Claude, Grok and 30+ models through one endpoint.
Base URL: https://jzstoken.com/api/v1
Supported Model Providers
Switch models per task — chat, reasoning, code, vision — without changing your client config.
General-purpose chat, content generation, translation, customer service.
DeepSeek · Qwen · Doubao · GPT · ClaudeDeep reasoning, document analysis, 1M-token context windows for book-length processing.
Kimi · GLM · StepFun · Claude OpusCode generation, tool calling, autonomous agent workflows, multi-step reasoning.
Qwen Coder · LongCat · MiMo · GPT CodexImage understanding, diagram analysis, OCR, image generation workflows.
Hunyuan · MiniMax · Doubao · GPT ImageHere's a realistic monthly task — so you know exactly what to expect.
Monthly volume
Input
2M
tokens
Output
1M
tokens
Actual cost varies by model multiplier and input/output ratio. Check the pricing table for real-time rates per model.
Chat Completions, Responses, and Messages are all supported. Check whether your client wants a full endpoint or just a Base URL, then copy the matching address below.
/Fits the widest range of clients
The general-purpose choice. Most SDKs, plugins, and third-party clients that speak the standard chat completions protocol can point straight at this endpoint.
https://Built for next-gen response calls
For clients and workflows that speak the Responses format, so agent traffic and everything else stay under one gateway and one balance.
https://For Messages-format clients
For clients that require the Anthropic Messages format, such as Claude Code and Cursor. Check your client's docs for whether it wants a Base URL or the full endpoint.
https:/Some clients ask for a Base URL instead of a complete address. In that case use the value below and let the client append the rest — pasting a full endpoint into a Base URL field makes the client append the path twice, and the call fails.
https:/From DeepSeek to Claude, GPT to Grok — switch models instantly without changing keys. Prices are USD per 1M tokens, billed per token. Reading from the cache is discounted; writing to the cache is billed at its own rate.
| Model | Provider | Input / 1M | Output / 1M | Context | Vision |
|---|---|---|---|---|---|
| claude-sonnet-4-6HOT | AI API Proxy | $0.038 | $0.192 | 1M | ✅ |
| LongCat-2.0 | Meituan / LongCat | $0.077 | $0.307 | 1M | — |
| hy3 | Tencent Hunyuan | $0.077 | $0.384 | 256K | — |
| MiniMax-M3 | MiniMax | $0.077 | $0.384 | 1M | ✅ |
| step-3.7-flash | StepFun | $0.077 | $0.384 | 256K | ✅ |
| claude-haiku-4-5-20251001 | Anthropic | $0.077 | $0.384 | 256K | ✅ |
| gpt-5.6-luna | OpenAI | $0.077 | $0.461 | 258K | ✅ |
| gpt-5.6-terra | OpenAI | $0.077 | $0.461 | 258K | ✅ |
| deepseek-v4-flash | DeepSeek | $0.148 | $0.739 | 1M | — |
| mimo-v2.5 | Xiaomi / MiMo | $0.154 | $0.768 | 1M | ✅ |
| doubao-seed-2.0-pro | ByteDance / Doubao | $0.230 | $1.152 | 128K | ✅ |
| doubao-seed-2.0-code | ByteDance / Doubao | $0.230 | $1.152 | 200K | ✅ |
| qwen3.6-plus | Alibaba / Qwen | $0.230 | $1.152 | 1M | ✅ |
| gpt-5.4 | OpenAI | $0.230 | $1.383 | 1M | ✅ |
| mimo-v2.5-pro | Xiaomi / MiMo | $0.307 | $1.536 | 1M | — |
| qwen3.7-plus | Alibaba / Qwen | $0.307 | $1.536 | 1M | ✅ |
| grok-4.5 | xAI / Grok | $0.384 | $1.920 | 500K | ✅ |
| glm-5.1 | Zhipu / Z.ai | $0.384 | $1.920 | 256K | ✅ |
| deepseek-v4-pro | DeepSeek | $0.443 | $2.216 | 1M | — |
| gpt-5.6-sol | OpenAI | $0.461 | $2.765 | 258K | ✅ |
| gpt-5.5 | OpenAI | $0.461 | $2.765 | 258K | ✅ |
| glm-5.2 | Zhipu / Z.ai | $0.615 | $3.072 | 1M | ✅ |
| qwen3.8-max | Alibaba / Qwen | $0.768 | $2.304 | 1M | ✅ |
| qwen3.7-max | Alibaba / Qwen | $0.768 | $3.841 | 1M | — |
| claude-sonnet-5 | Anthropic | $0.768 | $3.841 | 1M | ✅ |
| claude-opus-4-8 | Anthropic | $1.152 | $5.761 | 1M | ✅ |
| claude-opus-4-7 | Anthropic | $1.152 | $5.761 | 1M | ✅ |
| claude-opus-4-6 | Anthropic | $1.152 | $5.761 | 1M | ✅ |
| claude-opus-5 | Anthropic | $1.152 | $5.761 | 1M | ✅ |
| kimi-k3 | Moonshot / Kimi | $1.920 | $9.601 | 1M | ✅ |
| claude-fable-5 | Anthropic | $3.072 | $15.362 | 1M | ✅ |
Billed per image, not per token — the price is the same regardless of prompt length. Call POST /api/v1/images/generations.
| Model | Provider | Price / image |
|---|---|---|
| gpt-image-2 | OpenAI | $0.0192 |
| gpt-image-2-4k | OpenAI | $0.1920 |
Anonymized from real support conversations.
“Internal knowledge base connected seamlessly. The team only maintains one API config, and everyone keeps using their preferred clients.”
“Customer QA, product copy, and translation all share one Key. Model switching per task makes monthly cost tracking straightforward.”
“Long-doc extraction, code assist, and reasoning tasks need different models. Shared balance saved students from applying for multiple accounts.”
“Delivering AI workflows to clients with both OpenAI-compatible and Messages endpoints. Switching clients doesn't require redoing the entire config.”
Not just an API. Everything you need to use it in production.
We do not log request content. Your data is forwarded, never stored.
Contact support with your order number for official invoicing.
WhatsApp instant support. Topping-up issues, refunds — we handle it.
Common questions, answered upfront.
Register an account, top up credits, and create an API Key in your dashboard. Keys are instantly active.
Use https://jzstoken.com/api/v1 as your base URL. Append /chat/completions, /responses, or /messages depending on your protocol.
No. Credits never expire. Top up what you need, when you need it. No subscription, no minimum.
30+ models including DeepSeek, Qwen, Kimi, GLM, GPT, Claude, Grok, Doubao, Hunyuan, MiniMax, LongCat, StepFun, and MiMo. Full list on the pricing page.
Each model has a base price per 1M tokens and a multiplier. Your cost = basePrice × multiplier × tokens. Check the pricing table for real-time rates.
Check that your API Key is correct (no extra spaces), the Authorization header uses 'Bearer', and the base URL is correct. Contact support if it persists.
More questions? View Full FAQ or Contact Support