inference optimization

Liquid AI released DSpark draft model checkpoints for three LFM2.5 models, enabling speculative decoding that speeds up inference by up to 3.18x on H100 GPUs and 2.87x on-device, with day-one support for llama.cpp and SGLang.

OpenAI cut GPT-5.6 Luna pricing by 80% to $0.20/$1.20 per million tokens and GPT-5.6 Terra by 20% to $2/$12, while adding a 2.5x-faster mode for Sol at double the price. The company says GPT-5.6 itself rewrote production inference kernels and tuned its own speculative decoding pipeline to enable the cuts.

NVIDIA has released Nemotron-Labs-3-Puzzle-75B-A9B, a compressed variant of Nemotron-3-Super that reduces the model from 120.7B total/12.8B active parameters to 75.3B total/9.3B active parameters. According to NVIDIA, the model achieves approximately 2× higher server throughput on a single 8×B200 node and increases sustainable 1M-token single-H100 concurrency from 1 request to 8 requests while maintaining strong accuracy across benchmarks.

NVIDIA released Nemotron-Labs-Diffusion-14B, a 14-billion parameter language model that supports three decoding modes by switching attention patterns during inference. The model achieves 850 tokens per second on GB200 hardware at concurrency 1, representing a 3.3x speed-up over standard autoregressive decoding and outperforming Qwen3-8B-Eagle3 by 2.2x in self-speculation mode.