Inference API

Run any model. Keep your data in Europe.

OpenAI-compatible access to 60+ open-weight models on sovereign European infrastructure. One API key, automatic failover, and full transparency.

Get your API Key Read the Docs
Request
{
  "model": "deepseek-v3.2",
  "messages": [
    { "role": "user",
      "content": "Explain how GDPR applies to AI." }
  ]
}
Response
{
  "id": "chatcmpl-meli_7f3a...",
  "model": "deepseek-v3.2",
  "choices": [{
    "message": {
      "content": "The GDPR applies to..."
    },
    "finish_reason": "stop"
  }],
  "environment_impact": {
    "energy_kwh": 0.00057,
    "carbon_g_co2": 0.33,
    "water_liters": 0.00057,
    "location": "NL"
  }
}
Meta (Llama)
DeepSeek
Mistral AI
Qwen
Gemma
Nous Research
KIMI
OpenAI

Works with everything that works with OpenAI.

Change your base URL. Keep your code.

main.py
from openai import OpenAI

client = OpenAI(
    base_url="https://api.melious.ai/v1",
    api_key="your-api-key"
)

Drop in the Melious base URL and you're done. Every SDK, framework, and tool that speaks OpenAI works out of the box.

Same model. Different priorities.

Append a routing mode to any model ID. The router optimizes for your chosen target.

Default. Optimizes across price, speed, and environmental impact.

Just add a suffix
model: "deepseek-v3.2"

60+ models. One API.

From chat to image generation, embeddings to speech. All accessible through a single, consistent API.

Browse all models

Large Language Models

Chat completions with streaming, function calling, vision, and JSON mode.

Embeddings

Generate vector embeddings for search, RAG, and clustering.

Image Generation

Generate images from text with models like Flux.

Speech & Audio

Speech-to-text and text-to-speech, ready for production.

Reranking

Re-score and reorder search results for better relevance.

9+ providers. Automatic failover.

We built on a network, not a single server. If one provider goes down, your requests keep flowing.

Multi-provider network

9+ European inference providers across 8 countries. No single point of failure.

Automatic failover

If a provider is slow or down, requests automatically route to the next healthy one.

Real-time health monitoring

Every provider is continuously monitored for latency, errors, and availability.

Per-request carbon tracking. Because you should know.

0.57 Wh Energy
0.33g CO₂
0.57 ml Water

Every API response includes environmental optional impact data. Route with :eco to minimize it.

Built for developers.

A familiar API, great docs, and the tools you already know.

from openai import OpenAI

client = OpenAI(
    base_url="https://api.melious.ai/v1",
    api_key="your-api-key"
)

response = client.chat.completions.create(
    model="deepseek-v3.2",
    messages=[
        {"role": "user",
         "content": "Hello from Europe!"}
    ]
)

print(response.choices[0].message.content)

Streaming

Real-time, token-by-token output via server-sent events.

Function Calling

OpenAI-compatible tool use for agents and workflows.

Vision

Send images alongside text for multimodal understanding.

JSON Mode

Structured output with guaranteed valid JSON.

Batch API

Process thousands of requests asynchronously at lower cost.

Start free. No credit card required.

Get started.

Get your API Key See Pricing