Kolibri open-weight model launches with 78B parameters and 1M token context

Kolibri, a new open-weight AI model, has been released with 78 billion total parameters and 3 billion active parameters. The model supports a context window of up to 1 million tokens and is distributed under the Apache 2.0 open-source license, making it freely usable and modifiable.

The model is designed as a Mixture-of-Experts architecture, which uses selective activation of parameters to improve efficiency. It is marketed as suitable for sovereign and mission-critical applications, suggesting a focus on organizations that want to run AI workloads independently without reliance on proprietary platforms.

The release adds to a growing ecosystem of open-weight alternatives to closed models, giving developers and enterprises more options for building AI systems with full control over the underlying technology.