MiniStaff API docs
One key, one wallet, every model — at exactly half the official price. The API speaks the OpenAI chat-completions format and the Anthropic messages format, so your existing tools work by changing the base URL.
Quickstart
Get a key from your dashboard, top up at least $10, then call:
curl https://api.ministaff.pro/v1/chat/completions \
-H "Content-Type: application/json" \
-H "Authorization: Bearer msst_your_key_here" \
-d '{
"model": "gpt-6-luna",
"messages": [{"role": "user", "content": "Hello!"}]
}'
Add "stream": true to the same request body for a streaming response. The response shape matches OpenAI’s chat-completions format, including usage with prompt and completion tokens.
Authentication
Send your key on every request, either as a bearer token or as the x-api-key header:
Authorization: Bearer msst_...
# or
x-api-key: msst_...
Keys start with msst_. You can create up to 10 keys per account from the dashboard, and revoke any key at any time. Keep keys secret — never commit them to a repository or ship them in client-side code. If a key leaks, revoke it and create a new one.
Endpoints
| Method | Endpoint | What it does |
|---|---|---|
| POST | /v1/chat/completions | Chat completions, OpenAI-compatible (streaming supported) |
| POST | /v1/messages | Messages, Anthropic-compatible |
| POST | /v1/responses | Responses API, OpenAI-compatible (streaming supported) — what Codex uses |
| GET | /v1/models | List available models with their prices |
Base URL for all endpoints: https://api.ministaff.pro/v1
Models & prices
Prices are USD per 1M tokens, input / output — exactly half the official price. Workhorse models carry 1M-token context. Cached input is billed at 1/10 of the normal input rate.
| Model | Input / 1M | Output / 1M |
|---|---|---|
| DeepSeek V4.1 Flash | $0.075 | $0.30 |
| GLM 5.3 Flash | $0.075 | $0.25 |
| GLM 5.3 | $0.70 | $2.20 |
| Kimi K3 | $1.50 | $7.50 |
| GPT-6 Luna | $0.05 | $0.25 |
| GPT-6 Sol | $1.00 | $5.00 |
| GPT-6.1 Sol | $1.00 | $5.00 |
| GPT-6.1 Terra | $0.50 | $2.50 |
| GPT-5.6 Sol | $2.00 | $10.00 |
| GPT-6 Astra | $5.00 | $25.00 |
| Grok 4.7 | $1.00 | $3.00 |
| Claude Haiku 5.5 | $0.05 | $0.25 |
| Claude Sonnet 5.5 | $1.00 | $5.00 |
| Claude Opus 5.5 | $2.00 | $10.00 |
| Claude Fable 5.1 | $5.00 | $25.00 |
| Claude Fable 5.2 | $4.00 | $20.00 |
| Gemini 4 Argon | $1.00 | $5.00 |
| Gemini 4 Carbon | $1.00 | $5.00 |
| Gemini 3.8 Flash | $0.375 | $1.875 |
Billing
- Prepaid wallet. Top up $10–$100 by card, or by UPI (India: UPI only). Your dashboard itemises every call by model, tokens and cost.
- Bonus credit. +5% on $25+, +10% on $50+, +15% on $100+, as extra wallet credit. Per-token rates never change.
- Free starter credit. New accounts get $0.50 of promotional credit to try the API — one per account (and only the first account from a network), valid 30 days, on most models — not Kimi K3, GLM 5.3, Claude Fable 5.1 or Claude Fable 5.2. Promotional credit is spent before your paid balance.
- Refer & earn. Share your referral link from the dashboard: when a friend makes their first paid top-up of $25 or more, you both get $5 credit. Referral credit expires 30 days after it lands — see the Terms.
- Pause at $0. When your balance hits zero, keys pause. Top up and the same key works again.
- Refunds. Top-ups are non-refundable once credited — see the Refund Policy for the exceptions.
Limits
- 60 requests per minute per account. No daily cap.
- Up to 10 API keys per account.
- 1M-token context on workhorse models.
- No peak pricing. One flat rate per model, all day.
Errors
| Status | Error | What it means |
|---|---|---|
| 401 | invalid key | Missing, malformed or revoked API key. Check the key and try again. |
| 402 | insufficient_quota | Your API balance is empty. Top up to keep using the API. |
| 429 | rate_limit_error | Too many requests, or too many failed attempts. Wait a moment and retry. |
| 502 | proxy_error | The model service is unreachable right now. Retry shortly. |
Using it with your tools
Any tool that takes a custom base URL and key works — Claude Code, Cursor, Codex and most coding tools. For Claude Code and other Anthropic-compatible tools, set the base URL to the origin (the tool appends /v1/messages itself) and pick a model from the table above:
export ANTHROPIC_BASE_URL="https://api.ministaff.pro"
export ANTHROPIC_AUTH_TOKEN="msst_..."
export ANTHROPIC_MODEL="claude-sonnet-5-5"
For OpenAI-compatible tools, set the base URL to https://api.ministaff.pro/v1 and the API key to your msst_... key. Step-by-step: the Claude Code setup guide and the Cline setup guide.
Codex speaks the Responses API, which MiniStaff serves at POST /v1/responses. Add a provider block to ~/.codex/config.toml, set your key in an environment variable, and pick any model from the table above:
model = "deepseek-v4.1-flash"
model_provider = "ministaff"
[model_providers.ministaff]
name = "MiniStaff"
base_url = "https://api.ministaff.pro/v1"
wire_api = "responses"
env_key = "MINISTAFF_API_KEY"
export MINISTAFF_API_KEY="msst_..."
Support
Stuck? Write to [email protected]. The legal fine print lives in the Terms, Privacy Policy and Refund Policy.