Nemotron-Labs-Diffusion: A Tri-Mode Language Model Unifying Autoregressive, Diffusion, and Self-Speculation Decoding

NVIDIA released Nemotron-Labs-Diffusion, a tri-mode language model that unifies autoregressive, diffusion, and self-speculation decoding within a single architecture. The model, trained with a joint AR-diffusion objective, demonstrated that diffusion improves lookahead planning while AR provides left-to-right linguistic priors, and in self-speculation mode, diffusion drafting with AR verification outperformed multi-token prediction methods in acceptance rate and efficiency. Scaling from 3B to 14B parameters, the Nemotron-Labs-Diffusion family consistently outperformed state-of-the-art open-source models in accuracy and speed, with the 8B variant decoding 5.9× more tokens per forward pass than Qwen3-8B and achieving 4× higher throughput on SPEED-Bench.

We introduce Nemotron-Labs-Diffusion, a tri-mode language model LM that unifies AR, diffusion, and self-speculation decoding within a single architecture. Trained with a joint AR-diffusion objective, Nemotron-Labs-Diffusion can switch modes to sustain high throughput across deployment settings and concurrency levels. Our study shows that 1 AR and diffusion objectives are complementary: diffusion improves lookahead planning, while AR provides left-to-right linguistic priors. 2 In self-speculation mode, diffusion drafts while AR verifies, outperforming multi-token prediction MTP methods in both acceptance rate and real-device efficiency. 3 A speed-of-light analysis further demonstrates diffusion’s long-term potential, with up to 76.5% more tokens per forward pass than self-speculation under an optimal sampler. Scaling to 3B, 8B, and 14B parameters, our Nemotron-Labs-Diffusion family, including base, instruct, and vision-language models, consistently outperforms state-of-the-art open-source AR and diffusion LMs in both accuracy and speed. For example, Nemotron-Labs-Diffusion-8B decodes 5.9×more tokens per forward than Qwen3-8B with better accuracy, translating to 4× higher throughput on SPEED-Bench with SGLang on a GB200 GPU. HF collection: https://huggingface.co/collections/nvidia/nemotron-labs-diffusion https://huggingface.co/collections/nvidia/nemotron-labs-diffusion