Build with AI, without the concerns
One OpenAI-compatible API for 15+ open models, priced in euros. 2.5% of every request funds forest restoration.
https://platform.aim2balance.ai/api/a2b/v1Everything you expect, plus a planet that gets something back
- Drop-in compatible. Works with the OpenAI and Anthropic SDKs. Change the base URL and key, keep your code.
- One key, many models. Kimi, GLM, Qwen, Mistral, DeepSeek, MiniMax and Gemma. Switch with one parameter.
- Smart routing.
a2b-core2anda2b-boost2pick the right model for each step and run web search for you. - Prepaid, in euros. Top up a wallet and pay per token. A request your balance cannot cover is refused before it runs.
- Spend under control. Spending limits, alerts and allowed models on every key.
- Funds restoration. 2.5% of every request funds ecosystem restoration with Bergwaldprojekt e.V.
AI should work for us, not against us.
Four commitments behind every request you send.
Sustainable by design
- 80% less energy & water*
- 100% renewable energy
- 2.5% restoration fees per transaction
- Real-time footprint data
* Compared with conventional AI, based on Ecologits.
What a prompt really costsIndividual and national sovereignty
- EU hosting. Every prompt runs on servers inside the EU.
- GDPR compliance. Plus the EU AI Act, built in from the start.
- AES-256 encryption. Bank-grade protection for your data.
- Open models. Against power-concentration.
Serving humans, not investors
- AI-Balance Bar. See how you use AI at a glance.
- Mindful nudges. Gentle prompts to pause, never pressure.
- No engagement loops. No streaks, infinite feeds or notification bait.
- You stay in charge. Limits and reminders are yours to set.
Pay-as-you-go, with full transparency
- Pay per prompt. Prepaid credit, no subscription.
- Credit never expires. Use it today or next year.
- Every cent visible. See what each prompt costs.
- No investors. A non-profit, so no profit pressure.
Quickstart
Your key, the base URL and a first request in curl, Python or Node.
Your first request in three steps
- Create an account. Sign up and top up your wallet.
- Create an API key. Generate a key in the API dashboard and store it as an environment variable.
- Send a request. Use any OpenAI SDK with our base URL.
Authentication and keys
Send your key as a bearer token in the Authorization header. Create keys in Settings › API dashboard and keep them in an environment variable, never in client-side code.
Authorization: Bearer $A2B_API_KEYCode examples
The same request in curl, Python and JavaScript. Any OpenAI SDK works: change the base URL and key.
curl https://platform.aim2balance.ai/api/a2b/v1/chat/completions \
-H "Authorization: Bearer $A2B_API_KEY" \
-H "Content-Type: application/json" \
-d '{
"model": "a2b-core2",
"messages": [
{ "role": "user", "content": "Explain tidal energy in two sentences." }
]
}'import os
from openai import OpenAI
client = OpenAI(
base_url="https://platform.aim2balance.ai/api/a2b/v1",
api_key=os.environ["A2B_API_KEY"],
)
reply = client.chat.completions.create(
model="a2b-core2",
messages=[{ "role": "user", "content": "Explain tidal energy in two sentences." }],
)
print(reply.choices[0].message.content)import OpenAI from "openai";
const client = new OpenAI({
baseURL: "https://platform.aim2balance.ai/api/a2b/v1",
apiKey: process.env.A2B_API_KEY,
});
const reply = await client.chat.completions.create({
model: "a2b-core2",
messages: [{ role: "user", content: "Explain tidal energy in two sentences." }],
});
console.log(reply.choices[0].message.content);{
"model": "a2b-core2",
"choices": [{ "message": { ... } }],
"usage": {
"prompt_tokens": 18,
"completion_tokens": 52
}
}The API returns the answer and token usage. Energy, water and restoration are shown in the Usage tab of your API dashboard.
How it works
Personal account (individual users)
- Get access. Go to Settings > API dashboard > Open API dashboard, then top up your wallet to unlock the API.
- Create an API key. Open the Keys tab and click Create key. Copy the key straight away, because it's only shown once.
- Try it in the Playground. Ask for a long answer, then show the stats. The token count appears under the response. The Usage tab then shows spend, requests, tokens, and the environmental impact (energy, water and restoration).
Organization (org owners)
- Turn on API access. As the owner, open the organization dashboard, go to the API tab, and switch the API on under "API settings".
- Tour the API section. The same tab shows API spend, requests, active keys and the API vs chat split. Below that are the models used, member access requests, the members list with each person's keys and spend, and service keys for shared work.
Models
Pick a model or let us pick
- A2B routers.
a2b-core2anda2b-boost2choose a model for each step. Boost leans on the strongest models. 128k context. - A2B code.
a2b-codeis tuned for coding agents like Claude Code and Codex. 1M context. - Single models. Kimi K3, GLM 5.3, MiniMax M3, DeepSeek V4 Flash, Qwen 3.8 Max, Mistral Medium 3.5, Gemma 4 and more.
You can also call single models such as Kimi K3, GLM 5.3, MiniMax M3, DeepSeek V4 Flash, Qwen 3.8 Max or Mistral Medium 3.5 directly. See the price list for context sizes and rates.
Pricing and costs
Pay per token, in euros. No subscription. Top up a prepaid wallet and pay only for what you use.
Models and prices
Provider list prices per million tokens, before fees.
| Modell | ID | Typ | Kontext | Eingabe / 1M | Ausgabe / 1M |
|---|---|---|---|---|---|
| A2B-core2 | a2b-core2 | Router | 128k | 1,00 € | 3,00 € |
| A2B-boost2 | a2b-boost2 | Router | 128k | 3,00 € | 15,00 € |
| A2B-code | a2b-code | Coding | 1M | 0,50 € | 2,10 € |
| Kimi K3 | kimi-k3 | Modell | 1M | 3,00 € | 15,00 € |
| GLM 5.3 | glm-5.3 | Modell | 1M | 1,00 € | 3,00 € |
| MiniMax M3 | minimax-m3 | Modell | 1M | 0,40 € | 2,00 € |
| Qwen 3.8 (Max) | qwen-3.8-max | Modell | 256k | 2,30 € | 5,45 € |
| Mistral Medium 3.5 | mistral-medium-3.5 | Modell | 262k | 1,50 € | 7,50 € |
| DeepSeek V4 Flash | deepseek-v4-flash | Modell | 1M | 0,40 € | 0,80 € |
| GLM 5.3 Flash | glm-5.3-flash | Modell | 128k | 0,25 € | 0,50 € |
List prices from the model registry, before fees. Router runs also bill the router's own prompts, so a short message uses more input tokens than you sent. See every model in the reference.
How the price is made up
You pay per token from a prepaid wallet in euros:
- the model's list price,
- plus a 10% platform fee,
- plus 2.5% for ecosystem restoration,
- plus a 3% exchange margin for USD-priced models only.
A request your balance cannot cover is refused before it runs.
Stay in control of spend
- One wallet for chat and API. Top up once, use it everywhere. We hold the estimated cost while a request runs and charge only the actual cost. VAT may apply depending on where you are billed.
- Limits you set. Give each key a spending limit, an alert, a max tokens ceiling and a list of allowed models. Organizations can also give each member a daily, weekly or monthly budget.
Billing questions
What happens when my balance runs out?
The request is refused before the model runs, and nothing is charged.
Is there a monthly fee?
No. You pay per token from your prepaid balance.
Where does the 2.5% go?
To ecosystem restoration by Bergwaldprojekt e.V. Your Usage tab shows what you funded.
Can I cap spend per key?
Yes. Set a spending limit and an alert on any key in the API dashboard.
Requests
Request parameters
Chat completions follow the OpenAI format: pick a model, send messages, and set the usual options. The full parameter list is in the reference.
Streaming
Stream responses with the standard streaming option of your SDK. See the reference for details and examples.
Claude and Anthropic clients
The API also works with the Anthropic SDKs and with tools built on them. Change the base URL and key, and keep your code.
Responses API
The Responses API is covered in the reference, with request and response examples.
Embeddings
Create embeddings through the same base URL and key. See the reference for the available models.
Idempotency
Retry requests safely. How idempotency works is described in the reference.
Rate and concurrency limits
By default each key can make 120 requests per minute and run 8 requests at once.
| Limit | Standard, pro Schlüssel |
|---|---|
| Anfragen | 120 pro Minute |
| Gleichzeitige Anfragen | 8 |
API keys
Give each key a spending limit, an alert, a max tokens ceiling and a list of allowed models. Organizations can set a daily, weekly or monthly budget per member and create service keys that keep working when people leave.
Organizations
Built for organizations
One shared wallet. Admins decide who gets access, set a budget per member and create service keys that keep working when people leave.
The organization wallet shows the balance, a personal daily limit and recent transactions.
Client setup
Use it where you already work. Every tool here has a setup guide in the docs.
Setup guides for Claude Code, Codex, Cursor, Cline, aider, Continue, opencode, Open WebUI, LibreChat, LangChain, LlamaIndex and the Vercel AI SDK.
a2b-code or a single model.Errors
Errors use standard HTTP status codes with a JSON body. The reference lists every error and what to do about it.
Full reference
Models, parameters, limits, pricing, keys, streaming and errors, plus machine-readable files (openapi.json, llms.txt, Markdown for agents):