活跃 API KeyActive Keys
-
24h 请求数24h Requests
-
总 Token 用量Total Tokens
-
熔断状态Circuit Breaker
-
📊 24小时请求趋势24h Request Trends
📈 24小时 Token 用量24h Token Usage
⚡ 快速开始Quick Start
cURL
curl https://api.stormengine.cloud/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "Qwen2.5-32B",
"messages": [{"role": "user", "content": "你好"}],
"stream": false
}'
-H "Content-Type: application/json" \
-H "Authorization: Bearer YOUR_API_KEY" \
-d '{
"model": "Qwen2.5-32B",
"messages": [{"role": "user", "content": "你好"}],
"stream": false
}'
🔑 创建 API KeyCreate API Key （需管理员权限）(Admin required)
Key 列表Key List （每账号最多 2 个 Key）(Max 2 keys per account)
| 名称Name | ID | 用量 / 配额Usage / Quota | 限速Rate | 状态Status | 操作Actions |
|---|
今日费用Today (估算est.)
$0
本月累计This Month
$0
定价Pricing
$/M tokens$/M tokens
💵 各 Key 费用明细Per-Key Billing
| Key | Token 用量Token Usage | 今日费用Today | 本月估算Month Est. | 配额使用Quota Used |
|---|
📈 费用趋势 (24h)Cost Trends (24h)
💳 充值Recharge
充值通道暂未开放，请联系我们获取 API Key。Recharge not yet available. Contact us for API Key access.
💬 对话测试Chat Test
输入问题测试 STORM 模型Type a question to test STORM
📡 可用模型Available Models
Qwen2.5-32B DGX Spark
上下文Context: 64K tokens
最大输出Max Output: 8192 tokens
最大并发Max Concurrency: 30
平均延迟Avg Latency: ~19s
Function Calling: ✅
流式输出Streaming: ✅ SSE
JSON 模式Mode: ✅
Qwen2.5:14B Mac M4
最大并发Max Concurrency: 18
平均延迟Avg Latency: ~9s
用途Role: 轻量任务分流Lightweight offload
Qwen3-8B FALLBACK
上下文Context: 32K tokens
用途Role: 熔断降级Circuit-breaker fallback
🔗 API Endpoints
| POST | /v1/chat/completions | 对话补全 (兼容 OpenAI SDK)Chat Completions (OpenAI-compatible) |
| GET | /v1/models | 模型列表Model List |
| GET | /health | 健康检查Health Check |
| GET | /admin/usage | 用量查询 (需 API Key)Usage Query (API Key required) |
🐍 Python SDK 示例Python SDK Example
Python
from openai import OpenAI
client = OpenAI(
base_url="https://api.stormengine.cloud/v1",
api_key="YOUR_API_KEY"
)
response = client.chat.completions.create(
model="Qwen2.5-32B",
messages=[{"role": "user", "content": "你好，STORM！"}],
max_tokens=1024,
temperature=0
)
print(response.choices[0].message.content)
client = OpenAI(
base_url="https://api.stormengine.cloud/v1",
api_key="YOUR_API_KEY"
)
response = client.chat.completions.create(
model="Qwen2.5-32B",
messages=[{"role": "user", "content": "你好，STORM！"}],
max_tokens=1024,
temperature=0
)
print(response.choices[0].message.content)
📦 Node.js 示例Node.js Example
JavaScript
import OpenAI from 'openai';
const client = new OpenAI({
baseURL: 'https://api.stormengine.cloud/v1',
apiKey: 'YOUR_API_KEY'
});
const response = await client.chat.completions.create({
model: 'Qwen2.5-32B',
messages: [{ role: 'user', content: 'Hello STORM!' }]
});
console.log(response.choices[0].message.content);
const client = new OpenAI({
baseURL: 'https://api.stormengine.cloud/v1',
apiKey: 'YOUR_API_KEY'
});
const response = await client.chat.completions.create({
model: 'Qwen2.5-32B',
messages: [{ role: 'user', content: 'Hello STORM!' }]
});
console.log(response.choices[0].message.content);
🔐 认证方式Authentication
所有 API 请求需在All API requests require an API Key in the Authorization header 中携带 API Key：header:
Authorization: Bearer sk-storm-xxxxxxxx
⚡ STORM AI — Pricing
Built for developers who need determinism, dedicated capacity, and data sovereignty — not the cheapest tokens on the market. Pay for guaranteed throughput on physical hardware, not shared cloud quota.
Plans
Free
$0/mo
Try it out, side projects.
- 100K tokens/month
- 5 req/min
- Qwen2.5-32B
- Shared concurrency
- Community support
Experience
$3.9/mo
Getting started with Agents.
- 500K tokens/month
- 10 req/min
- Qwen2.5-32B
- Shared concurrency
- Community support
Starter
$9.9/mo
Testing an Agent in pre-production.
- 1M tokens/month
- 10 req/min
- Qwen2.5-32B + 14B
- Shared concurrency
- Community support
Developer
$49/mo
Indie devs running a single Agent.
- Unlimited tokens
- 20 req/min
- 1 dedicated slot
- 32B + 14B
- 48h email support
- 99% SLA
Most popular
Pro
$159/mo
Small teams, multiple Agents.
- Unlimited tokens
- 60 req/min
- 4 dedicated slots
- 32B + 14B
- 24h email support
- 99.5% SLA
Business
$799/mo
Production workloads, no limits.
- Unlimited tokens
- No rate limit
- 16 dedicated slots
- 32B + 14B
- 4h priority support
- 99.5% SLA
Enterprise
$4,999+/mo
Compliance, on-prem, custom.
- Custom capacity
- On-premise deploy
- Custom fine-tuning
- Your data residency
- 24/7 dedicated engineer
- 99.9% SLA
Full Feature Comparison
| Feature | Free | Experience | Starter | Developer | Pro | Business | Enterprise |
|---|---|---|---|---|---|---|---|
| Monthly tokens | 100K | 500K | 1M | Unlimited | Unlimited | Unlimited | Unlimited |
| Rate limit | 5/min | 10/min | 10/min | 20/min | 60/min | None | None |
| Dedicated concurrency slots | — | — | — | 1 | 4 | 16 | Custom |
| Available models | 32B | 32B | 32B + 14B | 32B + 14B | 32B + 14B | 32B + 14B | All + Custom |
| TTFT P50 | ~1.5s | ~1.5s | ~1.5s | ~1.5s | ~1.2s | ~1.0s | <1s |
| SLA uptime | — | — | — | 99% | 99.5% | 99.5% | 99.9% |
| Data residency | China | China | China | China | China | China | Your choice |
| Support response | Community | Community | Community | 48h | 24h | 4h | 24/7 |
| Custom system prompts | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Tool calling | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Streaming output | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Usage dashboard | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ | ✓ |
| Custom fine-tuning | — | — | — | — | — | — | ✓ |
| On-premise deployment | — | — | — | — | — | — | ✓ |
Pay-as-you-go (Alternative)
Prefer per-token billing instead of a subscription? You can charge usage to your account balance:
| Model | Input | Output | Best for |
|---|---|---|---|
| Qwen2.5-32B (STORM main) | $0.50 / 1M | $1.00 / 1M | Complex code, Agents, long context |
| Qwen2.5-14B (Mac M4) | $0.20 / 1M | $0.40 / 1M | Simple tasks, classification, extraction |
When to subscribe vs pay-as-you-go: Subscriptions become cheaper after roughly 3M tokens/month on the 32B model.
Most production Agents cross this threshold within days. If you run a real Agent, subscribe.
Why STORM vs OpenAI / Claude / DeepSeek
| Capability | STORM | OpenAI GPT-4o-mini | Anthropic Claude 3.5 | DeepSeek |
|---|---|---|---|---|
| Output price ($/1M tokens) | $1.00 | $0.60 | $15 | $0.30 |
| Dedicated capacity (no noisy neighbor) | ✓ | — | — | — |
| No rate limits on paid tier | ✓ | — | — | — |
| Data stays on our hardware (never trained on) | ✓ | Opt-out | Opt-out | Opt-out |
| Tool calling / Function calling | ✓ | ✓ | ✓ | ✓ |
| Full system prompt control | Full | Limited | Limited | Limited |
| 1,280 tests, 0 structural errors | ✓ | n/a | n/a | n/a |
Get Started in 60 Seconds
STORM is OpenAI API compatible. Drop in your key, change the base_url, and you're running.
# 1. Install the OpenAI SDK
pip install openai
# 2. Drop in your STORM key
from openai import OpenAI
client = OpenAI(
base_url="https://api.stormengine.cloud/v1",
api_key="YOUR_STORM_KEY",
)
response = client.chat.completions.create(
model="Qwen2.5-32B",
messages=[{"role": "user", "content": "Write a Python sort function."}],
temperature=0, # Deterministic output for Agent pipelines
)
print(response.choices[0].message.content)
100,000 free tokens every month. No credit card required for the Free tier.
Payment Methods
💳 Credit / Debit Card (Stripe)
🅿️ PayPal
🏦 Wire Transfer (Business+, $500+)
₿ USDC / USDT (on request)
FAQ
Why is STORM cheaper than OpenAI but more expensive than DeepSeek?
We're not a cloud giant with shared inference. Each paying customer gets dedicated capacity on physical hardware. You're paying for guaranteed throughput, not just tokens.
What happens when I hit my Free token cap?
Requests return HTTP 429 until the next month. Upgrade anytime to remove the cap immediately.
Can I cancel anytime?
Yes. Monthly subscriptions, cancel anytime in your dashboard. No questions, no retention calls.
Where is my data processed?
On our NVIDIA DGX Spark in Nanjing, China by default. Enterprise customers choose their deployment region.
Do you train on my data?
No. Ever. Prompts and outputs are never used for training. Logs are kept for billing and abuse-prevention only, and auto-purge after 30 days.
Refund policy?
7-day full refund on the first month of any paid plan. No questions asked.
How is the SLA calculated?
SLA = (Total time − Downtime) / Total time, measured monthly. Credits are issued automatically when we miss the target.
What's a "dedicated slot"?
A guaranteed concurrent inference slot on our hardware. You can have 1 (Developer) to 16 (Business) requests in flight without queuing — no noisy neighbors.
Enterprise & Custom
Need something beyond Business? We work directly with compliance-sensitive teams:
On-premise deployment
Install STORM on your hardware, behind your firewall.
Custom fine-tuning
Train Qwen2.5 on your domain data, hosted privately.
Air-gapped environments
Fully isolated deployments for classified workloads.
Compliance
SOC2, HIPAA, GDPR DPA — ask for our compliance pack.
Dedicated engineer
24/7 named contact, Slack channel, custom integrations.
Volume discounts
50%+ off list at high commitment levels.
Contact: [email protected]
Trust & Transparency
Open benchmark data
github.com/YIQI-NUMBER1/stormengine
Hardware
NVIDIA DGX Spark (verified) + Apple M4 Pro
Inference engine
vLLM 0.21 (open source)
Models
Qwen2.5 series (open weights)
Ready to build?
Get a free key in 60 seconds. No credit card.
At-a-Glance
Free$0/mo · 100K tokens
Starter$9/mo · 1M tokens
Developer$49/mo · Unlimited + 1 slot
Pro$159/mo · Unlimited + 4 slots
Business$799/mo · Unlimited + 16 slots
Enterprise$4,999+/mo · Custom + on-prem
📧 联系我们Contact Us
有建议、投诉或合作意向？直接留言，内容将发送至我们的邮箱。Suggestions, feedback, or partnerships? Leave a message and we'll get back to you.