HomeDocumentation

Documentation

One OpenAI-compatible endpoint for different models. Go from your first request to streaming responses, one step at a time.

API DOCUMENTATION · PREVIEWStart building

Quickstart

Four steps, from creating a key to reviewing usage. Choose cURL, Python or Node.js for the examples below.

/01

Create an account and key

Create an API key in the console and configure its models and quota.

Manage API keys

/02

Point your client at the API

Set the Base URL and read your Bearer key from the environment.

View Base URL

/03

Pick a model ID

Browse the catalogue and choose a model accessible to your key.

Choose a model

/04

Review your usage

Check requests and consumption in the usage dashboard.

Open usage dashboard

Base URL

https://www.tokiai.ai/v1

Set LEAPRIVER_KEY in your server environment first. Install the openai package for Python or the openai dependency for Node.js. Replace example/ model IDs with models available to your key.

curl https://www.tokiai.ai/v1/chat/completions \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "example/chat-model",
    "messages": [{"role": "user", "content": "Say hi in one line."}]
  }'

View example responseMOCK

View example response

{
  "id": "chatcmpl_example",
  "object": "chat.completion",
  "model": "example/chat-model",
  "choices": [
    {
      "index": 0,
      "message": {
        "role": "assistant",
        "content": "Hello! How can I help?"
      },
      "finish_reason": "stop"
    }
  ],
  "usage": {
    "prompt_tokens": 14,
    "completion_tokens": 8,
    "total_tokens": 22
  }
}

Authentication

Send a Bearer key in the Authorization header of each request. Use an API key, not your console login token.

curl https://www.tokiai.ai/v1/models \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json"

Keep keys in server-side environment variables, never in Git, browser code or shared messages. Use separate keys for development and production. Disable and replace any exposed key.

Chat completions

POST/chat/completions

Pass a conversation as an OpenAI-style messages array and receive an assistant message. Supported parameters depend on the model and the published API.

curl https://www.tokiai.ai/v1/chat/completions \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "example/chat-model",
    "messages": [
      {"role": "system", "content": "You are a concise assistant."},
      {"role": "user", "content": "Summarise this changelog."}
    ],
    "max_tokens": 400,
    "temperature": 0.2
  }'

Key parameters

Parameter	Type	Notes
`model`Required	string	Target model ID. Example IDs are not live models; select one from the available catalogue.
`messages`Required	array	Conversation messages with role and content. Supported roles depend on the model.
`max_tokens`	integer	Maximum output tokens, subject to the model's context and output limits.
`temperature`	number	Sampling temperature. Availability and valid range depend on the model.
`stream`	boolean	Use true for an SSE stream, otherwise receive the complete response.
`route`	string	Prototype extension: price / latency / reliability routing preference. Omitted from copyable requests.Backend pending
`fallback`	array	Prototype extension: backup models if the primary is unavailable. Omitted from copyable requests.Backend pending

Streaming

POST/chat/completions

Set stream to true to receive Server-Sent Events (SSE). Stop reading at [DONE] and handle empty choices or delta values.

curl -N https://www.tokiai.ai/v1/chat/completions \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "example/chat-model",
    "stream": true,
    "messages": [{"role": "user", "content": "Count to five."}]
  }'

View example responseMOCK

View example response

data: {"choices":[{"index":0,"delta":{"content":"One"}}]}

data: {"choices":[{"index":0,"delta":{"content":", two, three, four, five."}}]}

data: {"choices":[{"index":0,"delta":{},"finish_reason":"stop"}]}

data: [DONE]

Listing models

GET/models

List model IDs accessible to your key for your application's model selector. Extended pricing and context fields await backend confirmation.

curl https://www.tokiai.ai/v1/models \
  -H "Authorization: Bearer $LEAPRIVER_KEY"

Browse models →

View example responseMOCK

View example response

{
  "object": "list",
  "data": [
    {
      "id": "example/chat-model",
      "object": "model",
      "owned_by": "example"
    },
    {
      "id": "example/vision-model",
      "object": "model",
      "owned_by": "example"
    }
  ]
}

Vision input

POST/chat/completions

Combine text and image_url entries in the message content array. Vision-enabled models can receive image URLs or base64 data URIs.

curl https://www.tokiai.ai/v1/chat/completions \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "example/vision-model",
    "messages": [{
      "role": "user",
      "content": [
        {"type": "text", "text": "What does this chart show?"},
        {"type": "image_url", "image_url": {"url": "https://example.com/chart.png"}}
      ]
    }]
  }'

Image URLs must be accessible to the server; valid base64 data URIs may also be used. Vision support, multiple images and size limits depend on the selected model.

Image generation

POST/images/generations

The prototype's image endpoint returns image URLs or base64 data from a prompt. The example shows request syntax only; this page does not execute generation.

curl https://www.tokiai.ai/v1/images/generations \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "example/image-model",
    "prompt": "Studio product photo of an amber glass bottle.",
    "n": 1,
    "size": "1024x1024"
  }'

Video generation

POST/videos

The prototype uses an asynchronous task: submit to obtain an ID, then query its status. These proposed HTTP examples avoid unconfirmed SDK extensions.

# 1. Submit an example task (proposed API; backend pending).
curl https://www.tokiai.ai/v1/videos \
  -H "Authorization: Bearer $LEAPRIVER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"example/video-model","prompt":"A quiet forest.","seconds":8}'

# 2. Replace job_example with the ID returned by the server.
curl https://www.tokiai.ai/v1/videos/job_example \
  -H "Authorization: Bearer $LEAPRIVER_KEY"

This is the prototype's proposed asynchronous contract, not a connected generation service. Check queued / processing / completed / failed states after submission. SDK methods, polling policies and result URL expiry await backend confirmation.

View example responseMOCK

View example response

{
  "id": "job_example",
  "status": "queued",
  "model": "example/video-model"
}

Errors

Example OpenAI-style errors and common recovery strategies. Distinguish input, authentication and transient service errors rather than retrying every failure.

Status	Meaning	Action
`400`	Malformed request or invalid parameter	Check JSON, model ID and parameters. Do not retry unchanged.
`401`	Missing, invalid or disabled key	Check Authorization and key status before requesting again.
`429`	Rate or concurrency limit exceeded	Respect Retry-After and use bounded exponential backoff.
`502 / 503`	Upstream service temporarily unavailable	Limit retries; avoid duplicate tasks or charges for creation requests.
`529`	Prototype provider-overload status	Confirm published status codes. Retry later or choose an available model.

View example responseMOCK

View example response

{
  "error": {
    "message": "Rate limit exceeded. Please retry later.",
    "type": "rate_limit_error",
    "param": null,
    "code": "rate_limit_exceeded"
  }
}

Rate limits

Limits depend on the API key and model configuration. The headers below illustrate request counts, remaining tokens and retry timing.

Example response headers

x-ratelimit-limit-requests: 3000
x-ratelimit-remaining-requests: 2987
x-ratelimit-limit-tokens: 400000000
x-ratelimit-remaining-tokens: 391204556
retry-after: 5

These mock values are not actual quotas or an SLA. Respect Retry-After and use exponential backoff with a retry limit. Do not retry indefinitely.

Documentation

Create an account and key

Point your client at the API

Pick a model ID

Review your usage

Base URL

Key parameters

Start with your first API request