Inference (Prev. Kuzco) is a company that builds and hosts custom AI models tailored to specific use cases. It offers services such as:
Model training and design
Inference serving (serverless or dedicated)
Batch processing
Data extraction and document understanding
The platform supports multiple modalities (text, image, video, audio) and structured output generation. It also provides infrastructure and tooling so that teams can deploy and scale models reliably.
Inference.net positions itself as an alternative to large “frontier” models by optimizing for lower latency, cost, and predictable performance, while avoiding vendor lock-in and hidden changes to model behavior.
