Leuks — Foundation Models
Leuks is a family of Foundation AI Models designed and developed by Inferent. These models power AI experiences across our product ecosystem and are available to developers and enterprises through our API.
We are committed to developing AI that is capable, safe, and aligned with human values. The documents in this section provide full transparency into how our models work, what they are designed to do, and the rules governing their use.
Responsible AI Policy
Our principles and commitments to safe, fair, and beneficial AI.
Model Card
Training methodology, capabilities, limitations, and known biases.
Usage Policy
What you can and cannot do with Leuks models.
Content Policy
Safe, restricted, and prohibited content categories.
API Terms
Terms governing API access, rate limits, and data handling.
AI Safety Policy
Red teaming, alignment, and incident response.
Transparency Report
Content removals, safety incidents, and model updates.
About Leuks Models
Leuks models are large language models (LLMs) trained on diverse, carefully curated data sources. Our training methodology follows modern, responsible AI development practices and includes:
- Publicly available text from the internet and digitized books, filtered for quality and safety
- Licensed datasets from data providers who have agreed to our data use policies
- Internally generated synthetic datasets designed to improve specific capabilities and safety behaviors
- Reinforcement Learning from Human Feedback (RLHF) to align model outputs with human preferences
- Dedicated safety optimization pipelines to reduce harmful, biased, and misleading outputs
We do not train our models by copying copyrighted works without authorization. Our approach is conservative, transparent, and continuously evolving as the field of AI safety advances.