API for
developers.

A2Intelligence speaks the OpenAI Responses API. Existing SDKs work right away – you only change the base URL, API key and model name.

Base URL: https://gateway.a2genesis.de/v1
Model: a2intelligence
Auth: Authorization: Bearer $A2_API_KEY

first-request · gateway.a2genesis.de/v1/responses

1curl https://gateway.a2genesis.de/v1/responses \2  -H "Authorization: Bearer $A2_API_KEY" \3  -H "Content-Type: application/json" \4  -d '{5    "model": "a2intelligence",6    "input": "Erkläre die DSGVO in drei Sätzen."7  }'

request.shPOST /v1/responses

completedsimulated

response.output_text

The GDPR governs how companies and public bodies in the EU may process personal data. Every processing activity needs a legal basis, and data subjects have rights such as access, rectification and erasure. Violations can be fined.

input_tokens: 18
output_tokens: 56
status: completed

content-type: text/event-stream41 events

Get started in three steps.

From account to first answer.

Step 01
Create an account
Sign up on the platform and top up credit or choose a subscription.
platform.a2genesis.de
Step 02
Create an API key
Generate an API key in your workspace – as many as you need, one per application.
export A2_API_KEY=…
Step 03
Send a request
Use the base URL https://gateway.a2genesis.de/v1 and the model a2intelligence.
Go to the endpoints

Endpoints.

Three endpoints under https://gateway.a2genesis.de/v1. For new projects we recommend the Responses API.

Headers

Authorization: Bearer $A2_API_KEY

Content-Type: application/json

https://gateway.a2genesis.de/v1

POST

/v1/responses

The standard interface: instead of a list of messages you pass an input. Reasoning is kept separate from the answer, and tool calls are structured.

Parameters (body)

Name	Type	Description
modelrequired	string	"a2intelligence"
inputrequired	string \| array	Text or a list of inputs – including images for image understanding.
instructions	string	Instructions for the model, such as role or tone.
stream	boolean	Stream the answer as server-sent events.
reasoning	object	Reasoning effort, e.g. { "effort": "high" }.
tools	array	Functions the model may call in a structured way.
max_output_tokens	integer	Upper limit for the length of the answer.

RequestJSON

{
  "model": "a2intelligence",
  "instructions": "Antworte knapp.",
  "input": "Erkläre die DSGVO in drei Sätzen.",
  "reasoning": { "effort": "medium" },
  "stream": true
}

Response (shortened)JSON

{
  "object": "response",
  "status": "completed",
  "model": "a2intelligence",
  "output": [ … ],
  "output_text": "Die DSGVO regelt …",
  "usage": { "input_tokens": …, "output_tokens": … }
}

POST

/v1/chat/completions

The /v1/chat/completions endpoint remains supported so existing integrations keep working unchanged. New projects should use the Responses API.

Parameters (body)

Name	Type	Description
modelrequired	string	"a2intelligence"
messagesrequired	array	Conversation made of roles (system, user, assistant) and content.

curlLegacy example

curl https://gateway.a2genesis.de/v1/chat/completions \
  -H "Authorization: Bearer $A2_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "a2intelligence", "messages": [{"role": "user", "content": "Hallo"}]}'

Moving to the Responses API: messages becomes input, the answer is in output_text.

GET

/v1/models

Lists the models available with your API key. Useful as a connection test and for tools that detect models automatically.

Parameters

None – only the Authorization header.

Requestcurl

curl https://gateway.a2genesis.de/v1/models \
  -H "Authorization: Bearer $A2_API_KEY"

Response (excerpt)JSON

{
  "object": "list",
  "data": [{
    "id": "a2intelligence",
    "object": "model"
  }]
}

Change three lines.
Keep the rest.

Existing OpenAI SDKs work. You only change the base URL, API key and model name.

 import OpenAI from "openai"
  
 const client = new OpenAI({
+  baseURL: "https://gateway.a2genesis.de/v1",
−  apiKey: process.env.OPENAI_API_KEY,
+  apiKey: process.env.A2_API_KEY,
 })
  
 const response = await client.responses.create({
−  model: "<previous-model>",
+  model: "a2intelligence",
   input: "Erkläre die DSGVO in drei Sätzen.",
 })

Official OpenAI SDKs

For JavaScript/TypeScript and Python – no separate SDK needed.

$ npm i openai$ pip install openai

Streaming via SSE

With "stream": true you receive events in the Responses format:

response.output_text.delta
response.reasoning_summary_text.delta
response.completed

Reasoning, tools and images

All capabilities of A2Intelligence through the same interface – up to 131,072 tokens of context per request.

See capabilities

Operated in Germany
Not at US hyperscalers.
No training on your data
Input is only used for your request.
Billed in euros
Pay-as-you-go credit or subscription.

Create an API key
and get started.

You can also try A2Intelligence without code in the platform playground.

Start the platform View pricing

From Chat Completions to the Responses API FAQ Discuss an integration

API for
developers.

Get started in three steps.

Create an account

Create an API key

Send a request

Endpoints.

/v1/responses

/v1/chat/completions

/v1/models

Change three lines.
Keep the rest.

Official OpenAI SDKs

Streaming via SSE

Reasoning, tools and images

Create an API key
and get started.

Read more

A2Intelligence – the German LLM

LLM API pricing and plans

AI developer platform by A2Genesis

From Chat Completions to the Responses API

Switch from the OpenAI API to A2Intelligence

API fordevelopers.

Get started in three steps.

Create an account

Create an API key

Send a request

Endpoints.

/v1/responses

/v1/chat/completions

/v1/models

Change three lines.Keep the rest.

Official OpenAI SDKs

Streaming via SSE

Reasoning, tools and images

Create an API keyand get started.

Read more

A2Intelligence – the German LLM

LLM API pricing and plans

AI developer platform by A2Genesis

From Chat Completions to the Responses API

Switch from the OpenAI API to A2Intelligence

API for
developers.

Change three lines.
Keep the rest.

Create an API key
and get started.