Liquid AI Releases Open d1 Decision Models for Multimodal Edge Inference

Liquid AI Releases Open d1 Decision Models for Multimodal Edge Inference Liquid AI has announced two open-weight decision models, d1-3B and d1-omni-600M, built on Liquid Foundation Models to execute fast, structured multimodal decisions in a single forward pass without token generation. d1-3B scored 48.57 on Decision Index 0.2.1, surpassing competing models under 10B as well as Decider 35B-A3B. d1-3B is trained from the decoder-only LFM2.5-VL-3B, while d1-omni-600M adds vision and audio encoders to the bidirectional LFM2.5-Encoder-350M backbone. On seven public language and classification benchmarks, d1-3B achieved a mean score of 82.9, while d1-omni-600M scored 78.4. Single-question inference for d1-3B runs in 16 ms on NVIDIA Jetson AGX Thor, 50 ms on Jetson Orin Nano, and under 10 ms on desktop GPUs. Model Architectures and Benchmark Performance Unlike generative LLMs, the d1 decision models deliver structured outputs in a single forward pass. The d1-3B model utilizes a decoder-only LFM2.5-VL-3B backbone handling text and image inputs. The experimental d1-omni-600M uses a bidirectional LFM2.5-Encoder-350M backbone augmented with vision and audio encoders, accepting text alongside either images or audio. Across seven public evaluation benchmarks—covering reading comprehension, toxicity detection, intent classification, medical QA, and cross-lingual understanding—d1-3B reached an average score of 82.9, outperforming Decider 4B. The 600M model reached an average of 78.4, exceeding the larger Decider 2B. Inference Latency on Edge and GPU Hardware Evaluations conducted with NVIDIA demonstrated low-latency execution for d1-3B. On edge hardware, single-question latency ranged from 16 ms on the Jetson AGX Thor to 50 ms on the Jetson Orin Nano, with processing scaling efficiently to 20 ms and 73 ms respectively when handling three questions simultaneously. On discrete GPUs such as the NVIDIA RTX 4090 and AMD MI325X, single-question queries resolved in 8 ms and 9 ms, with 384px image processing taking under 18 ms. Deployment and Availability Both open-weight models are hosted on Hugging Face and can be loaded via the Transformers library using trust_remote_code=True. Inference calls utilize the system_one method to evaluate multiple structured questions—such as boolean flags, choices, and scores—or packed inputs without padding across single text states and image inputs.

Liquid AI开源边缘多模态决策模型d1-3B与d1-omni-600M Liquid AI发布了两款开源边缘决策模型d1-3B与d1-omni-600M，通过单次前向传播而非生成Token的方式提供快速结构化决策支持，并在边缘设备与GPU上展现出极低的推理延迟。 模型基于LFMs构建，不生成文本Token，而是通过单次前向传播输出结构化决策结果。 d1-3B在Decision Index 0.2.1上得分48.57，位列10B以下模型第一且超越Decider 35B-A3B。 d1-3B支持文本和图像输入，d1-omni-600M进一步融合视觉和音频编码器，支持文本+图像或文本+音频。 在NVIDIA Jetson AGX Thor上，d1-3B单问题回答仅需16毫秒；在RTX 4090 GPU上低于10毫秒。 模型架构与多模态特性 两款模型基于Liquid Foundation Models（LFMs）构建，具有完全不同的骨干网络。d1-3B基于仅解码器的LFM2.5-VL-3B训练，支持文本与图像输入；d1-omni-600M则基于双向编码器LFM2.5-Encoder-350M构建，增加了视觉和音频编码器，支持文本与图像或文本与音频输入，目前属于早期研究版本。 性能评测与推理延迟 在涵盖阅读理解、毒性检测与多语言理解等7个公开数据集的测试中，d1-3B取得82.9的平均分，超越Decider 4B；d1-omni-600M以四分之一的参数量取得78.4分，超越Decider 2B（77.1）。在推理速度方面，d1-3B在边缘设备Jetson AGX Thor上处理单个问题仅需16毫秒，在RTX 4090和AMD MI325X GPU上单问题延迟分别仅为8毫秒和9毫秒。 开源部署与使用方式 两款模型均为开放权重，已在Hugging Face上提供下载。用户需安装transformers（>=5.14）及PyTorch等依赖，通过AutoModel配合trust_remote_code=True加载模型，支持对单文本状态进行多命名问题查询、图像状态分类以及无填充的批量请求处理。

Today, we release two open decision models in our d1 decision model family : d1-3B and d1-omni-600M (experimental). Best decision model under 10B on the Decision Index 0.2.1: d1-3B scores 48.57, ahead of every 4B and 9B model and of Decider 35B-A3B (47.11). Multimodal: d1-3B supports text and images, while d1-omni-600M supports text and images or text and audio Fast: d1-3B answers a question in 16 ms on an NVIDIA Jetson AGX Thor, 26 ms on a Jetson AGX Orin, and 50ms on a Jetson Orin Nano How we built decision models for the edge These open d1 decision models are built on our Liquid Foundation Models (LFMs). Unlike our generative models, decision models don’t produce tokens but answer in a single forward pass. d1-3B and d1-omni-600M are trained from two very different backbones: d1-3B is trained from LFM2.5-VL-3B , our latest VLM, which is decoder-only. It accepts text and images as inputs. d1-omni-600M is trained from LFM2.5-Encoder-350M , a bidirectional encoder. It adds vision and audio encoders to handle all three modalities. It accepts either text and image, or text and audio as inputs. This model is currently in an early research release and is undergoing further development. Benchmark results We benchmarked d1-3B and d1-omni-600M on seven public datasets spanning reading comprehension, toxicity detection, intent classification, medical QA, and cross-lingual understanding. d1-3B achieves a mean score of 82.9, the highest in the table and above Decider 4B. d1-omni-600M scores 78.4, surpassing Decider 2B (77.1) with only a quarter of the parameters. Benchmark d1-omni-600M d1-3B Decider 2B Decider 4B SQuAD 2.0 74.0 83.3 67.7 76.0 Civil Comments 95.8 93.3 93.6 92.8 MASSIVE intent 86.1 86.9 81.1 88.3 PubMedQA 61.3 68.3 65.7 63.3 BoolQ 77.7 86.3 87.3 89.0 XNLI 74.7 85.6 85.0 88.6 PAWS-X 79.5 76.4 59.5 69.8 Mean 78.4 82.9 77.1 81.1 We validated that d1-3B retains the vision capabilities of its LFM2.5-VL-3B backbone on standard vision benchmarks, and that d1-omni-600M handles all three modalities. We do not report any vision or audio benchmarks, as the Decision Index v0.3 includes only a private vision split and audio decision benchmarks are currently an open problem. Speed In collaboration with NVIDIA, we evaluated d1-3B on the NVIDIA stack across NVIDIA GeForce RTX 4090, NVIDIA Jetson AGX Thor, Jetson AGX Orin 64 GB, and Jetson Orin Nano. Since d1-omni-600M is an early research release, we don’t report any speed numbers for it in this release. Edge inference. d1-3B answers a single question in under 50 ms on every measured device. Three questions take only 1.3x the time of one, with the AGX Thor going from 16 ms to 20 ms. One question 3 questions 3.4K-token state 384px image 64 states, packed Apple M5 Pro 30 ms 41 ms 640 ms 62 ms 78 / s Jetson AGX Thor 16 ms 20 ms 220 ms 35 ms 262 / s Jetson AGX Orin 64 GB 26 ms 35 ms 560 ms 83 ms 110 / s Jetson Orin Nano 50 ms 73 ms 1,640 ms 202 ms 38 / s GPU inference. On GPU, d1-3B answers a question in under 10 ms and processes a 384px image in under 18 ms on both platforms. One question 3 questions 3.4K-token state 384px image 64 states, packed NVIDIA RTX 4090 8 ms 21 ms 102 ms 17 ms 475 / s AMD MI325X 9 ms 14 ms 44 ms 18 ms 1,106 / s How to use open d1 decision models Reach for d1 decision models when you need fast, structured decisions, including multimodal inputs. d1-3B delivers the highest decision quality at its size, while d1-omni-600M fits where footprint matters. Install the dependencies (requires transformers>=5.14 ): pip install "transformers>=5.14" torch torchvision pillow These model ship their own code, so load it with trust_remote_code=True : import io import urllib.request import torch from PIL import Image from transformers import AutoModel device = "cuda" if torch.cuda.is_available() else "mps" if torch.backends.mps.is_available() else "cpu" model = AutoModel.from_pretrained( "LiquidAI/d1-3B" , trust_remote_code= True , dtype=torch.float32 if device == "cpu" else torch.bfloat16).to(device) # Several named questions over one text state, answered in one pass questions = { "refund" : { "type" : "noul" , "instructions" : "Is the customer asking for a refund?" }, "team" : { "type" : "choice" , "instructions" : "Which team should handle this?" , "criteria" : { "billing" : "Charges, refunds, invoices" , "technical" : "App or site faults" , "fraud" : "Suspected unauthorised use" }}, "urgency" : { "type" : "score" , "instructions" : "How urgent is this?" , "criteria" : [ "Can wait" , "Today" , "Blocking the customer now" ]}, } print (model.system_one( "I was charged twice this month, please refund one of them." , questions)) # An image as the whole state url = "http://images.cocodataset.org/val2017/000000039769.jpg" # two cats on a sofa photo = Image. open (io.BytesIO(urllib.request.urlopen(url).read())) print (model.system_one( None , { "cats" : { "type" : "choice" , "instructions" : "How many cats are there?" , "criteria" : { "one" : "One" , "two" : "Two" , "more" : "Three or more" }}}, images=[photo])) # Many requests, packed together with no padding tickets = [ "Where is my parcel? It was due Monday." , "The app crashes when I open settings." ] print (model.system_one_batch([(t, { "team" : questions[ "team" ]}) for t in tickets])) For brevity, we only include the example for d1-3B. See the d1-omni-600M model card for instructions on how to run it. Get Started with open d1 decision models Both decision models are open-weight and available on Hugging Face today: Download: d1-3B and d1-omni-600M on Hugging Face. Try: run the demos in our System One Arcade Hugging Face Space. We can't wait to see what you build. Citation If you use this work, please cite the release blog: @article{liquidAI2026opend1, author = {Liquid AI}, title = {Open d1: Edge decision models for text, vision, and audio}, journal = {Liquid AI Blog}, year = {2026}, note = {www.liquid.ai/blog/open-d1}, }