Model Detail

Laguna-XS.2-NVFP4 is a code generation model with 9.8B parameters released by poolside. The model is registered under the text-generation pipeline tag on Hugging Face, distributed under the permissive apache-2.0 license.

Laguna-XS.2-NVFP4 ships with 9.8B parameters. Total weight footprint is approximately 19.6 GB, which is the relevant figure when planning local-inference VRAM. The apache-2.0 license is permissive, allowing commercial deployment and derivative work without per-seat fees, though attribution requirements still apply.

Laguna-XS.2-NVFP4 is best fit for code completion, repository-scale Q&A, and pair-programming integrations. It is a less obvious choice for one-shot generation of security-critical code without review. Treat this as a starting matrix rather than a benchmark verdict — the right deployment usually depends on the specific evaluation suite that mirrors your workload.