AI infrastructure for Africa

Open models.More control.

A simpler way to run open-weight AI. Start with an API, understand how it’s performing, and work with us on the deployment your workload needs.

Join early access Talk about your workload

Early access · Open for pilot conversations

From your app to the model

Your application

An OpenAI-compatible request

Tessil

API access · Routing · Request visibility

Open-weight model

Choose your model location

  • Lagos
  • Cape Town
  • Anywhere
  • Your data center
Open-weight modelsOpenAI-compatible APIRequest-level visibility

Built around your requirements

There’s more to AI
than a model.

Where it runs. How it performs. Who keeps it working. We’re building Tessil for African teams that need clear answers to those questions.

01 / ACCESS

Start with a familiar API.

Connect to an open-weight text model using the tools you already know. Create a key, choose a model, and send your first request.

02 / VISIBILITY

See what happened.

Track request outcomes and latency in your dashboard. Operational records leave out raw prompts and generated responses.

03 / DEPLOYMENT

Plan for the control you need.

Explore dedicated capacity or a deployment in your data center. Tell us about your workload, processing requirements, and budget to shape a pilot.

For developers

A small change
to your stack.

Keep your OpenAI-compatible client. Choose Lagos, Cape Town, or Automatic (Anywhere) in your dashboard, then use the provided base URL, model ID, and examples.

Text chat, streaming responses, and request history are available in the pilot product.

Open the early-access dashboard
Your first requestPython
import os
from openai import OpenAI

client = OpenAI(
    api_key=os.environ["TESSIL_API_KEY"],
    base_url=os.environ["TESSIL_BASE_URL"],
)

response = client.chat.completions.create(
    model=os.environ["TESSIL_MODEL"],
    messages=[{
        "role": "user",
        "content": "Hello, Tessil.",
    }],
)

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

Use the key, base URL, and model from your dashboard. Pilot access is currently by invitation.

Let’s start with one workload

What’s holding
your AI back?

Document workflows · Customer support · Internal tools

If infrastructure, cost, or data-handling requirements are slowing you down, we’d like to understand the problem.

Tell us what you’re building and what needs to work better. Together, we can scope a focused pilot and agree on what success looks like.

Let’s talk