NVIDIA: Nemotron 3 Nano 30B A3B (free)

Nemotron 3 Nano 30B A3B is a text-in, text-out model from NVIDIA with a 256,000-token context window. It supports tool use and reasoning, making it capable of multi-step and agentic workflows. Structured output support is unconfirmed, so developers who depend on guaranteed JSON or schema-constrained responses should verify that separately before committing to it. On the comparison side, the model is free with no input or output costs, which makes it a low-risk option for experimentation, prototyping, or cost-sensitive deployments. The catch is that independent benchmark coverage is entirely absent at this point, so there is no external data to compare its quality against competing models. Users should treat it as unproven and run their own task-specific evaluations before relying on it in production.