LFM2.5-1.2B-Instruct is a 1.2B-parameter instruction-tuned language model designed for strong capability at small scale and efficient on-device/edge deployment.
It builds on the LFM2 architecture (introduced July 2025) and improves performance by extending pretraining from 10T → 28T tokens and scaling post-training with reinforcement learning, targeting better instruction following and general usefulness in real-world tasks.
Text prompts or conversation history (tokenized sequences for APIs).
Output:
Generated text (answers, explanations, summaries, and structured content).
Optionally, raw logits/probabilities for advanced tasks.
License
This repo is licensed under the Creative Commons Attribution–NonCommercial 4.0 (CC BY-NC 4.0) license, which allows use, sharing, and modification only for non-commercial purposes with proper attribution. All NPU-related models, runtimes, and code in this project are protected under this non-commercial license and cannot be used in any commercial or revenue-generating applications. Commercial licensing or enterprise usage requires a separate agreement. For inquiries, please contact hello@nexa.ai
