Tokfai integration docs

AI API relay for developers and small teams. Path: sign up → top up ¥99 → create a key → set Base URL → call /v1/models → run chat → check Usage / Credits. Base URL: https://api.tokfai.com/v1

Early public beta — start small. Email junpengpanchina@gmail.com for a human review if integration, billing, image jobs, or usage logs look wrong.

¥99 API trial pack Integration path

Real integration path

Start setup

API integration

OpenAI-compatible endpoints: /v1/models, /v1/chat/completions, /v1/responses, /v1/images/generations.

View API docs

Client configuration

Connect Codex, Cursor, Cherry Studio, and similar tools to Tokfai.

View client setup

Real integration path

Authorization: Bearer sk-tokfai_xxx. Top up ¥99, verify models and chat, check Usage / Credits, then wire clients.

1Sign up / sign in, then top up the ¥99 trial pack
2Create a sk-tokfai_… API key; set Base URL to https://api.tokfai.com/v1
3Call /v1/models with curl, then /v1/chat/completions
4Check Usage / Credits for charges and failed requests, then wire Cherry Studio / Cursor

Base URL

https://api.tokfai.com/v1

GET /v1/models

curl -sS https://api.tokfai.com/v1/models \
  -H "Authorization: Bearer sk-tokfai_xxx"

POST /v1/chat/completions · stream=false

curl -sS https://api.tokfai.com/v1/chat/completions \
  -H "Authorization: Bearer sk-tokfai_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "auto-fast",
    "messages": [{"role":"user","content":"Reply with TOKFAI_OK"}],
    "stream": false
  }'

POST /v1/chat/completions · stream=true

curl -sS https://api.tokfai.com/v1/chat/completions \
  -H "Authorization: Bearer sk-tokfai_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "auto-fast",
    "messages": [{"role":"user","content":"Reply with TOKFAI_OK"}],
    "stream": true
  }'

Python · OpenAI SDK

from openai import OpenAI

client = OpenAI(
    api_key="sk-tokfai_xxx",
    base_url="https://api.tokfai.com/v1",
)

resp = client.chat.completions.create(
    model="auto-fast",
    messages=[{"role": "user", "content": "Reply with TOKFAI_OK"}],
)
print(resp.choices[0].message.content)

Node.js · OpenAI SDK

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "sk-tokfai_xxx",
  baseURL: "https://api.tokfai.com/v1",
});

const resp = await client.chat.completions.create({
  model: "auto-fast",
  messages: [{ role: "user", content: "Reply with TOKFAI_OK" }],
});
console.log(resp.choices[0].message.content);

Engineering material analysis

Tokfai cannot read local paths on your computer. Upload real files or a project archive.

UAV engineering material analysis

Upload UAV project archives or related materials. Tokfai reads content and outputs a diagnosis report.

1Upload materials
2Extract file content
3Model analysis
4Output markdown diagnosis report

Supported formats

zip, PDF, txt, md, log, json, csv, c, cpp, h, py, m, and more

Output file

tokfai-uav-diagnosis-YYYYMMDD-HHMMSS.md

Analysis scope

Flight approval
Airspace restrictions
Flight-control code
Attitude control
Task scheduling
Control allocation
Motor / throttle output
Safety boundaries
Log diagnostics

Script example

export TOKFAI_API_KEY="sk-tokfai_xxx"
python3 scripts/aviation/tokfai-uav-intake.py \
  --path "/你的工程目录" \
  --question "请分析无人机工程材料，重点看审批、空域、安全边界、飞控、姿态环、油门/电机输出和交付风险"

Why can't Tokfai read a path you type?

Local paths exist only on your machine — Tokfai in the cloud cannot read them. Sending a path like Flight/FlightControl/txg_control_task.c delivers text, not file content. Upload files via the intake script or upload API.

API integration

Authorization: Bearer sk-tokfai_xxx

—	https://api.tokfai.com	Base URL
GET	/v1/models	List available models
POST	/v1/chat/completions	Chat completions
POST	/v1/responses	Reasoning / Codex-style clients
POST	/v1/images/generations	Image generation

Client configuration

Connect Codex, Cursor, and Cherry Studio to Tokfai API.

Codex

config.toml

model = "gpt-5.5"
model_provider = "openai"
openai_base_url = "https://api.tokfai.com/v1"

auth.json

~/.codex/auth.json

{
  "OPENAI_API_KEY": "sk-tokfai_xxx"
}

Test

curl -sS https://api.tokfai.com/v1/models

curl -sS https://api.tokfai.com/v1/responses \
  -H "Authorization: Bearer sk-tokfai_xxx" \
  -H "Content-Type: application/json" \
  -d '{"model":"gpt-5.5","input":"回复 TOKFAI_AUTH_OK","stream":false}'

Common errors

`Missing Bearer token`	No Authorization header
`invalid_token`	API key wrong or expired
`404`	Wrong base_url — use https://api.tokfai.com/v1
`stream disconnected`	Client or proxy interrupted — disable proxy and retry
`model_not_available`	Model id not in Tokfai model list

Cursor

Connection

Base URL: https://api.tokfai.com/v1
API Key: sk-tokfai_xxx
Model: gpt-5.5 or a model id from Models

Local files

Whether Cursor analyzes local files depends on it sending file content in the request. Tokfai only processes what reaches the API.

Cherry Studio

Connection

Provider: OpenAI Compatible
Base URL: https://api.tokfai.com/v1
API Key: sk-tokfai_xxx
Fetch models from /v1/models

High-frequency calls & production

Tokfai is in early public beta: start with ¥99 to verify the API, latency, and cost before you scale.

Self-serve packs fit developers, small teams, Cherry Studio / Cursor, and mid-scale use — not formal large-buyer SLA procurement.

Million-RPM, distillation, training-data supply, large-scale scraping, grey-market / illegal use, and personal privacy / sensitive data are not self-serve offers.

If integration, billing, or usage logs look wrong, email junpengpanchina@gmail.com for a human review. Calls follow rate limits, Usage / Credits, and actual responses.

View pricing & scale notes