# AliAI

> AliAI is an AI gateway at https://aliai.alione.cc. It exposes one API key and one base URL for
> OpenAI-, Anthropic- and Gemini-compatible chat completions, routing each request to a model from
> Claude, OpenAI, Gemini, DeepSeek, Kimi, GLM or MiniMax. Usage is billed in tokens from the account's
> monthly allowance.

AliAI is a service by AliOne (https://alione.cc), contact support@alione.cc. It is a paid hosted
service, not a library. There is no public source repository for the gateway itself; the API surface
follows the public OpenAI, Anthropic and Google specifications.

## Base URL

https://aliai.alione.cc

## Authentication

- OpenAI-compatible: `Authorization: Bearer sk-...`
- Anthropic Messages: `x-api-key: sk-...` (Bearer also accepted)
- Gemini: `x-goog-api-key: sk-...` or `?key=sk-...`

Keys are created in the panel at https://aliai.alione.cc (API keys page) and shown once. Keys are
hashed at rest. If the account has not accepted the current Terms of Service, every API request
returns HTTP 403 with code `terms_not_accepted` until the owner signs in and accepts them.

## Endpoints

### POST /v1/chat/completions
OpenAI Chat Completions format. Body: `model`, `messages`, `temperature`, `top_p`, `max_tokens`,
`stop`, `stream`. Response: standard `chat.completion` object with `choices` and `usage`.

### POST /v1/messages
Anthropic Messages format. Body: `model`, `system`, `messages`, `max_tokens`, `temperature`, `top_p`,
`stop_sequences`, `stream`. Response: `{id, type: "message", role: "assistant", content: [{type:
"text", text}], stop_reason, usage: {input_tokens, output_tokens}}`. Streaming emits the standard
events: `message_start`, `content_block_start`, `content_block_delta`, `content_block_stop`,
`message_delta`, `message_stop`.

### POST /v1beta/models/{model}:generateContent
### POST /v1beta/models/{model}:streamGenerateContent
Gemini generateContent format. Body: `contents`, `systemInstruction`,
`generationConfig{temperature, topP, maxOutputTokens, stopSequences}`. Response: `candidates[]` with
`content.parts[].text`, `finishReason`, and `usageMetadata{promptTokenCount, candidatesTokenCount,
totalTokenCount}`.

### GET /v1/models
Lists the model ids available to the caller. Requires authentication.

## Streaming

All three protocols support server-sent events. OpenAI and Anthropic use `text/event-stream`;
Gemini uses `data: {json}` chunks. Streaming calls do not currently report prompt tokens in billing.

## Models

Model ids are AliAI public ids. The upstream provider and its internal model name are never exposed
in responses. Current public ids include:

- claude-opus-5-5, claude-fable-5-2, claude-sonnet-5-5, claude-haiku-5-5
- gpt-6-astra, gpt-6.1-sol
- gemini-4-argon
- deepseek-v4.1-flash, deepseek-ai-deepseek-v4-flash-0731
- kimi-k3.1-pro, kimi-k3
- hy4

Use GET /v1/models for the authoritative list; ids and availability change.

## Plans and billing

- Free: $0/month, 1,000,000 tokens
- Go: $5/month, 10,000,000 tokens
- Pro: $10/month, 50,000,000 tokens
- Ultra: $20/month, 100,000,000 tokens

Tokens are deducted per request from input plus output tokens. Balances and per-request charges are
visible in the panel. Top-ups and new payments are currently closed.

## Errors

- OpenAI shape: `{"error": {"message", "type", "code"}}`
- Anthropic shape: `{"type": "error", "error": {"type", "message"}}`
- Gemini shape: `{"error": {"code", "message", "status"}}`

Status codes: 400 invalid request, 401 missing or invalid key, 403 plan/quota/terms, 429 rate limit,
502 upstream failure.

## Limits and unsupported features

- Text only: no tool/function calling, no image or PDF input.
- No GET /v1beta/models listing.
- Context length depends on the selected model.
- Daily request and token limits may apply per key or account; they are shown in the panel.

## Links

- Documentation: https://aliai.alione.cc/docs
- Terms of Service: https://aliai.alione.cc/terms
- Privacy Policy: https://aliai.alione.cc/privacy
- Acceptable Use Policy: https://aliai.alione.cc/acceptable-use
- Cookie notice: https://aliai.alione.cc/cookies
- AI output disclaimer: https://aliai.alione.cc/ai-disclaimer
- Company: https://alione.cc
- Contact: support@alione.cc