Yerevan, Armenia  UTC+4

[email protected] +374 43 736370

Request a demo

DEVELOPERS

From API key to first answer in five minutes.

The AREVIA API follows the OpenAI Chat Completions format, so your existing SDKs, frameworks and tools work after changing two lines: the base URL and the model name.

01QUICKSTART

Three steps

  1. 01

    Create a key

    Sign in to AREVIA Studio, open “API keys” and create a key for your project. Keys can be limited to one environment.

  2. 02

    Set the base URL

    Point your client to https://api.areviaai.net/v1 and send the key as a Bearer token.

  3. 03

    Send a request

    Call chat completions with a model such as sevan-2. Set “stream” to true for token-⁠by-⁠token output.

curl https://api.areviaai.net/v1/chat/completions \
  -H "Authorization: Bearer $AREVIA_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "sevan-2",
    "messages": [{"role": "user", "content": "Summarise this contract in three bullet points."}]
  }'

200 OKResponse

{
  "id": "chatcmpl-7f3a9c2e",
  "model": "sevan-2",
  "choices": [{
    "index": 0,
    "message": {
      "role": "assistant",
      "content": "1. Term: 36 months from the date of signing.\n2. Rate: 11.5% per year, fixed.\n3. Early repayment is allowed without penalty."
    },
    "finish_reason": "stop"
  }],
  "usage": { "prompt_tokens": 1843, "completion_tokens": 96 }
}

02CAPABILITIES

What the API supports

  1. 01

    Chat completions

    Multi-⁠turn conversations with system instructions, in the languages your users speak.

  2. 02

    Streaming

    Token-⁠by-⁠token output over server-⁠sent events for responsive interfaces.

  3. 03

    Function calling

    The model calls your internal APIs with validated, typed arguments.

  4. 04

    Structured output

    Valid JSON that matches the schema you provide, every time.

  5. 05

    Embeddings

    Cross-⁠lingual vectors for search and retrieval with Garni.

  6. 06

    Batch processing

    Large asynchronous jobs at half the price of real-⁠time requests.

03REFERENCE

Endpoints

BASE URLhttps://api.areviaai.net/v1

METHOD PATH DESCRIPTION
POST /v1/chat/completions Generate a response from a language model
POST /v1/embeddings Create vector representations of text
POST /v1/audio/transcriptions Transcribe audio to text
POST /v1/audio/speech Synthesise speech from text
POST /v1/moderations Check content and mask personal data
POST /v1/batches Submit an asynchronous batch job
GET /v1/models List the models available to your key

Default rate limits

PLAN REQUESTS PER MINUTE TOKENS PER MINUTE BATCH QUEUE, TOKENS
Developer 60 200,000 1,000,000
Business 600 2,000,000 50,000,000
Enterprise Custom Custom Custom

Limits are raised on request, usually within one business day.

TOOLING

Works with what you already use

Official OpenAI SDKs for Python and JavaScript, LangChain, LlamaIndex and any HTTP client. No proprietary SDK is required: the PHP example on this page uses nothing but cURL.

NEXT STEP

Let’s put AI to work in your business.

Tell us about your process. Within five working days we prepare a pilot plan: scope, success metrics, timeline and a fixed price.