embirolabs
Join waitlist

Your stack.
Closer compute.

We’re designing an OpenAI-compatible interface for open models on African infrastructure.

Explore the API design Integration preview · no API keys yet
Open model compute layers.

Your next idea.
One familiar API.

Build with open models through a planned OpenAI-compatible API.

OpenAI-compatible APIDesign target for the pilot
POST/v1/chat/completions

Messages in. Model-generated text out.

POST/v1/embeddings

Vectors for search and retrieval.

GET/v1/models

Discover the available model catalogue.

Proposed routes. Compatibility and model IDs will be confirmed before the pilot.

Join the API waitlist
Embiro Labs / APINot live

Scroll sideways to explore the code

# Proposed interface — not a live endpoint
import os
from openai import OpenAI
 
client = OpenAI(
    base_url=os.environ["EMBIRO_BASE_URL"],
    api_key=os.environ["EMBIRO_API_KEY"],
)
 
response = client.chat.completions.create(
    model="glm-5-3",  # proposed model ID
    messages=[{
        "role": "user",
        "content": "Summarise this delivery note."
    }],
)
print(response.choices[0].message.content)

Integration preview. No live endpoint or API keys yet.

Your tools.
A closer engine.

Keep the workflows you know.
We’re targeting a familiar API.

Compatibility targets · validation before pilot

Independent tools, not Embiro partners. Their own hosting, storage and tracing can affect end-to-end data residency.

Familiar interface.
Clear boundaries.

The compatibility target is an OpenAI-style base URL, API key and model identifier. Final streaming, tool-calling, error handling and modality support will be documented and tested for the pilot.

We’ll publish working examples when the service is ready. Today’s code is a design preview, not production documentation.

Build closer. Be here first.

Tell us what you need from African AI infrastructure.

Join the waitlist No payment or commitment