On-Device Language Model Inference

Follow this topic Ask about this topic

Momentum

23 papers in the last four weeks, up 229% on the four weeks before. 0.2% of all new papers.

Jul 13Week of Sep 28

Latest papers 111

Oct 7, 2026cs.CL

Decoupling Logic from Persona: Structural Immunity of Edge LLM Agents to Context Pollution

Small language-model agents on edge devices must hold a persona and reason correctly at once, inside one context window that fills with conversational history and persona instructions. We study what happens to the logical part of such an agent when that history is long, misleading and persona-heavy (persona-logic interference), and present a Decoupling Architecture (AO-DA) that separates logical inference ("What") from persona expression ("How") into two inference paths on one INT4 base model with hot-swappable LoRA adapters. The logic path receives only the core turn and emits a verifiable structured state (Micro-State); the persona path renders it in character with the full history. In same-base-model ablations on an Apple M2 laptop (Llama-3.1-8B-Instruct and Gemma-3-4B-it, 4-bit; 480 runs over 4 pollution levels x 3 arms x 2 tasks x 2 personas x 5 seeds) we find: (i) the decoupled logic path is structurally invariant to pollution: its prompt stays at 180 (Llama) or 167 (Gemma) tokens while the mixed single-pass prompt grows from 242 to 1,203, and its outputs are byte-identical across levels (40/40); (ii) the mixed single pass degrades monotonically (composite logic score 0.669 to 0.150 on Llama, 0.487 to 0.150 on Gemma), mostly by failing to emit the required structured output (80-95% of runs on Llama, 100% on Gemma at the two highest levels); (iii) with the same pollution fed into the decoupled logic path, the dedicated-adapter, dedicated-format path is still more robust than the single pass on the 8B model (failure 0-20% vs 80-95%; paired

Δ

+0.30 to +0.50, Cliff's

δ

0.50-0.85, Holm-adjusted

p \le 0.03

) but not on the 4B model, where both collapse. Separation costs one extra decode on a topic's first turn (28.2 s vs 18.2 s on Llama) and buys persona hot-swapping in 1.7 ms without re-running the logic path. Code, rubric, fixtures, adapters and logs are released.

Masaaki Nakatsu, Reno Wang

AO, Inc. / OrbLabs AG · AO, Inc.

LLM Inference LLM Context Management On-Device Language Model Inference Persona-Conditioned Generation

Oct 5, 2026cs.LG

Stepped MoE: Segment-Level Routing with Configurable Inference Complexity

Training large language models (LLMs) is resource-intensive, and adapting them for diverse deployment scenarios with varying computational constraints remains challenging. While elastic architectures enable flexible model deployment and sparsely activated models allow input-adaptive computation, existing approaches treat these dimensions independently. Moreover, models catered towards on-device edge inference need to conform to the memory and compute limitations of the serving devices. In this paper, we introduce a unified framework that combines elastic structures with sparsely gated architectures to create models that adapt simultaneously to both deployment constraints and task requirements. Our approach employs a model backbone that conditions on both the context and target efficiency specifications, enabling fine-grained control over the accuracy-efficiency trade-off at inference time. The model learns to activate task-relevant parameters within elastically-nested sub-networks, allowing a single model to span multiple capacity points while maintaining input-adaptive routing. Through experiments we demonstrate that we can create a model that allows the flexibility to use 1,2,3,4 billion parameters while being more accurate than their dense counter-parts (2-5% on knowledge-intensive benchmarks) and at par with their static versions while delivering similar latency metrics as dense models. Overall, we save on device disk space by sharing the model parameters, allow flexibility of serving based on DRAM and compute available while delivering more accurate results.

Arnav Kundu, Zhaoyang Xu, Bairu Hou +3

Apple

On-Device Language Model Inference Dynamic Neural Networks Mixture of Experts LLM Routing Efficient Language Model Inference

Sep 30, 2026cs.IT

AIR-LLM: Broadcasting AI Weights over Radio for Memory-Free Edge LLM Inference via RF Computing

Next-generation large language models (LLMs) are expanding from the cloud to ubiquitous edge devices. However, edge devices typically either lack the memory to store increasingly large LLM weights or, even with enough memory, spend unaffordable energy on loading the weights. This raises our question: can an edge device run an LLM without storing or loading its weights, but receive them over the air and consume them on the fly? Inspired by wireless broadcasting, we present AIR-LLM, an LLM inference architecture for edge devices, which is composed of: (i) a central radio (e.g., 5G base stations) that broadcasts the LLM weights into the air, and (ii) the edge user that receives the weights and completes the general matrix-vector multiplication (GEMV) of LLM inference directly in the radio frequency (RF) domain using RF mixers. To further shorten the airtime, AIR-LLM exploits MIMO spatial multiplexing and proposes an energy-efficient precoder-postcoder pair on the edge to calibrate its own wireless channel. Since the central radio stays user-unaware, AIR-LLM is user-scalable so that one broadcast serves unlimited users within its coverage. We implement AIR-LLM on the NVIDIA Sionna ray-traced channels of two real-world urban scenes and the profiling of a real RF mixer. With a WikiText-2 perplexity degradation of 4.0% on LLaMA-3.1-8B, AIR-LLM saves the energy by 157.7x/40.4x against the FP16 and weight-only quantization baselines; with 20 users, its airtime is 104.1x/26.0x shorter, respectively.

Zhihui Gao, Tingjun Chen, Dirk Englund

Massachusetts Institute of Technology Cambridge, MA, USA · Duke University Durham, NC, USA

On-Device Language Model Inference LLM Inference Acceleration Wireless Communications

Sep 29, 2026cs.AI

HARISSA: Inference-Time Self-Checks for Efficient and Safe Local Language Model Deployment

Running a language model locally offers advantages in privacy, latency, and cost, but local hardware fits only small models, which are less capable than frontier models. The usual remedy for a hard query, escalating it to a cloud model, gives up the privacy and cost advantages of running locally. A deployment that stays local faces two decisions for hard queries instead. First, it can spend more computation on a query, e.g., reasoning before answering, which raises accuracy at a cost in latency, so it must decide which queries are worth the extra computation (efficiency). Second, some queries are beyond the local model, and delivering a wrong answer is worse than deferring the query to a human in the loop, so it must decide which answers are safe to deliver (safety). We show that both decisions can be made from the model's own hidden states. The prefill state, computed before any token is generated, predicts whether the model will answer correctly, and the answer state, at the end of the generated answer, predicts whether that answer is correct. HARISSA fine-tunes the model so that both states predict correctness, then makes both decisions with one policy that cascades through the ways of answering from cheapest to most expensive, skipping a way the prefill state predicts will fail and deferring the query when the answer it stops with is predicted wrong. On a device running a single model, HARISSA is within one accuracy point of chain-of-thought at 2.7 times lower latency. On a server holding four sizes of one model, HARISSA is more accurate than the FrugalGPT and Self-REF cascades at the same latency, and at the same deferral rate the answer state leaves fewer wrong answers than the standard confidence signals in five of six task and setting pairs.

Kenan Alkiek, Moontae Lee, David Jurgens +1

School of Information University of Michigan Ann Arbor, MI 48109, USA · SNU-LG AI Research Center Seoul, South Korea

On-Device Language Model Inference Selective Prediction Efficient Language Model Reasoning LLM Uncertainty Estimation Efficient Language Model Inference

Sep 29, 2026cs.SD

AS $^2$ D: Accelerating On-Demand Audio Understanding on Mobile Devices

Speculative decoding accelerates autoregressive generation by using a smaller drafter to propose tokens for batched verification by a larger target. However, conventional speculative decoding couples drafting to the target's evolving verified prefix, serializing drafting and verification. We ask whether this dependency is necessary for source-conditioned generation. Our key observation is that, for audio language models, the input audio and user request can provide useful speculative candidates without following the target's evolving text prefix. We propose AS

^2

D (Audio Speculative Speculative Decoding), which enables target-decoupled drafting: an audio-conditioned drafter follows its own generation history while the target independently verifies and corrects ready candidates. Without usable candidates, the target advances alone. Thus, target feedback determines which candidates are committed but no longer determines when the drafter can make progress, enabling drafting and verification to proceed concurrently while retaining target-side verification and correction. We implement AS

^2

D in MNN for Android and evaluate two target models across four phones, seven datasets, and three tasks covering 12.2 hours of audio. Across four phones, AS

^2

D improves pooled ASR throughput by 42-76% over target-only decoding, while only 5.7% of evaluation windows are slower than target-only, compared with 58.1-63.0% for speculative baselines. For ASR, AS

^2

D reaches 97.33-98.20% of a hindsight per-window oracle's pooled throughput over the evaluated drafter/budget catalog. Native on-demand execution with a 7B target achieves up to 78% higher throughput than target-only. These results show that source-conditioned audio generation can relax the conventional dependence of speculative drafting on the target's evolving output prefix, exposing substantial parallelism for efficient inference.

Yunzhe Li, Kyoungjun Park, Hongzi Zhu +1

The University of Texas at Austin · Shanghai Jiao Tong University

On-Device Language Model Inference Language Model Decoding Speculative Decoding Audio-Language Models Automatic Speech Recognition

Sep 29, 2026cs.AI

IronLLM: Forging Compact Edge-Native Language Models for Real-Time Embodied Intelligence

We present IronLLM-0.6B, a 654M-parameter language model designed for efficient on-device inference. IronLLM-0.6B combines a hybrid attention architecture with X-MTP, a lightweight shared-KV multi-token prediction design that eliminates per-depth KV-cache replay and employs a lightweight verification head for rollback-free drafting, achieving a 1.48x decoding speedup. The model is pretrained on approximately 6.2 trillion tokens using a quality-oriented data pipeline and is further post-trained with Multi-Domain On-Policy Distillation to integrate capabilities from domain-specialized teachers. To better meet the low-latency requirements of on-device scenarios, IronLLM-0.6B adopts an Instruct-Only design. Evaluations show that IronLLM-0.6B achieves competitive performance relative to larger models such as Qwen3.5-0.8B and MiniCPM5-1B, while producing more concise responses on many tasks. We further present IronLLM-0.6B-Light, which replaces RMSNorm with Dynamic Tanh and simplifies several computationally expensive components to improve inference and quantization efficiency. Together, the IronLLM models provide an effective performance-efficiency trade-off for resource-constrained deployment.

Changdi Yang, Fengquan Jiao, Haochih Lin +14

Robotics Foundation Model Team, Xpeng Inc.

Multi-Token Prediction On-Device Language Model Inference Small Language Models Speculative Decoding Efficient Language Model Inference

Sep 28, 2026cs.OS

Dynamic Flow, Static Graph: KV Cache Reuse for Efficient LLM Serving on Mobile NPUs

On-device large language model (LLM) serving is a cornerstone of local-first personal intelligence, offering users data sovereignty, strong privacy guarantees, and freedom from cloud API latency and cost. Although KV caching is widely used to reduce latency in long-context inference, existing designs were primarily optimized for cloud GPUs with dynamic execution environments and abundant memory bandwidth. These architectural assumptions do not hold on mobile NPUs, where computation graphs must be statically compiled and both memory capacity and I/O bandwidth are severely constrained. In this work, we present a compute-storage co-design for mobile-centric prefix and non-prefix KV reuse. We first propose an intra-graph mechanism that maps selective KV recomputation onto static NPU graphs, reconciling algorithmic dynamicity with NPU staticity. We further develop an inter-graph scheduler to optimize chunk merging and minimize padding with dynamic programming. To address mobile bandwidth limitations, we introduce a hierarchical KV manager featuring a tree-hash-semantic hybrid structure, along with cost-aware prefetching and eviction policies. We also build a two-dimensional pipeline that overlaps KV loading, rerotation, and storage with NPU execution, hiding data-movement latency. Experiments across representative on-device workloads and LLMs show that our design reduces time-to-first-token (TTFT) by

40-60\%

compared with no reuse and prefix-only caching.

Zhengxiang Huang, Shengheng Chen, Chaoyue Niu +6

Shanghai Jiao Tong University Shanghai, China · Alibaba Group Hangzhou, China · Alibaba Group Hangzhou, Zhejiang, China

AI Accelerator Inference On-Device Language Model Inference KV-Cache Management LLM Inference Acceleration Prefix Caching KV-Cache Reuse

Sep 28, 2026cs.AI

OmniTide: Co-Designing Algorithms and Systems for Efficient On-Device Omni-LLM Streaming

On-device streaming omni-modal inference safeguards user privacy and eliminates prohibitive per-token API costs, but faces a critical bottleneck: the continuous influx of multimodal data rapidly exhausts constrained memory and compute budgets via monotonic KV cache growth. Existing sparse attention methods fall short, either incurring prohibitive online estimation latency or destroying interleaved cross-modal context, while failing to resolve physical memory fragmentation. We present OmniTide, the first algorithm-system co-design tailored for efficient on-device streaming omni-modal inference. Driven by the observation of modality-aware structural sparsity, OmniTide adopts a unit-based abstraction with two components: (1) At the algorithm level, OmniPick logically retains critical multimodal context based on unit boundaries and modality importance to preserve task accuracy; (2) At the system level, OmniPage physically partitions the cache by retention likelihood and dynamically compacts surviving sparse tokens, minimizing both memory fragmentation and data-movement overhead. Extensive evaluations across three streaming benchmarks and two consumer-device architectures show that OmniTide achieves up to

12.72\times

kernel speedups and

2.40\times

lower stream-loop latency. On StreamingBench, it improves accuracy by up to 18.0 percentage points over sliding-window baselines at comparable session cost. OmniPage further reduces the physical KV span by up to 26.7% relative to native logical eviction, unlocking real-time, infinite-context streaming on edge devices.

Zongshang Shen, Wangsong Yin, Daliang Xu +2

Key Lab of High Confidence Software Technologies (Peking University), Beijing, China · State Key Laboratory of Networking and Switching Technology (BUPT), Beijing, China

Efficient Multimodal Inference On-Device Language Model Inference Multimodal Large Language Models KV-Cache Management

Sep 27, 2026cs.LG

Optimizing the Phi-2 Small Language Model for Real-time Chatbot Applications Using Parameter-Efficient Fine-Tuning (PEFT) with QLoRA Quantization

This study explores the optimization of the Phi-2 Small Language Models (SLMs) for real-time chatbot applications through Parameter-Efficient Fine-Tuning (PEFT) and Quantized Low-Rank Adaptation (QLoRA). QLoRA specifically refers to the integration of PEFT with LoRA alongside a 4-bit quantization process, aimed at enhancing computational efficiency. These models, initially designed for high performance with minimal computational overhead, are further refined to address the constraints of mobile and edge computing environments. By integrating PEFT with QLoRA, the research aims to reduce memory usage significantly while maintaining, or potentially improving, the accuracy of model responses in real-time interactions. The effectiveness of these techniques was evaluated using the ROUGE metric system, which showed notable improvements in the summarization tasks performed by the models. This approach not only confirms the feasibility of using SLMs in resource-restricted environments but also opens up new avenues for deploying advanced AI-driven applications in real-time settings. The study's findings have significant implications for the development of efficient, scalable, and accessible AI technologies, paving the way for broader adoption in various industries.

PhanTan Khanh Nguyen, Ashfaq Ali Shafin, Khandaker Mamun Ahmed

Beacom College of Computer and Cyber Sciences, Dakota State University · Mathematics and Computer Science, Augustana College

On-Device Language Model Inference Small Language Models Quantization-Aware LoRA Parameter-Efficient Fine-Tuning

Sep 24, 2026cs.PF

Paging the Experts: A Reproducible Characterization of Flash-Backed MoE Inference on iPhone

Sparse activation reduces mixture-of-experts computation without eliminating the need to store all experts. We present Routide, a Swift/MLX runtime that executes the text path of a pinned public Qwen3.6-35B-A3B quantized checkpoint while keeping expert weights in iPhone storage and a byte-budgeted subset in memory. We characterize cache-policy sensitivity, numerical comparison boundaries, and measurement limits. Across five recorded 128-token workloads, fixed-route replay gives 0.00% demand hits with a 512 MiB LRU cache, 18.80% with seeded random eviction at the same budget, and 38.58% with 576 MiB LRU. The apparent capacity cliff is therefore a policy/workload interaction, not a universal memory requirement. Same-runtime Mac controls preserve generated sequences across eviction and asynchronous prefetch, including 2,560 exact token comparisons and 10,334 speculative loads. In contrast, complete resident-Python versus recorded-phone sequences disagree on all five tested cases, precluding a general numerical equivalence claim. Two separately scoped iOS 27 memory protocols observe sampled process-footprint peaks of 1.87-2.32 GiB on short prompts and 2.39-2.73 GiB on one longer prompt. We retain a thermal stopping event, negative timing comparisons, and a single qualified whole-device power estimate. These results establish bounded feasibility and identify limitations that a deployment claim must not hide.

Musa Shams

Independent Researcher

Expert Offloading On-Device Language Model Inference Mixture-of-Experts Inference Memory-Efficient Inference

Sep 21, 2026cs.NE

Universal Fractal Natural Language Decision Map: Real-Time Edge Triage Across Heterogeneous Domains

Deploying Large Language Models for runtime operational triage incurs prohibitive latency (>100-500 ms), high VRAM requirements (>4-8 GB), and excessive energy dissipation. Extending Mandelbrot Fractal Neural Synthesis (Dagli et al., 2026), this paper presents the Universal Fractal Natural Language Decision Map, realized via the werr machine-native edge reflex runtime and the production answerr platform (https://answerr.me). Operating entirely without stored weight tensors (0 Bytes VRAM), the engine synthesizes deterministic decisions---noul (Boolean), choice (categorical), and score (ordinal)---by dynamically modulating 24-byte coordinate seeds along the chaotic boundary of the Mandelbrot set and evaluating multi-scale escape dynamics. Drawing inspiration from biological System-One reflex arcs, the engine introduces: (i) an Auto-Seed Router with domain projector Phi_D yielding a +28.8% accuracy gain over linear baselines; (ii) an Information-Theoretic Semantic Token Damping Filter (T_desc = 0.045) insulating against prompt injections (0.0% empirical bypass; 95% Wilson CI: [0.0%, 27.8%]) while pruning iterations by 45.8% (accelerating throughput 2.5x to 3.31 ms latency); (iii) a Multi-Scale Harmonic Tripod Fusion; (iv) a Coupled Margin Expansion Operator (Pitchfork Bifurcation Offset); and (v) a Cyclic Z/9Z Modular Resonant Grid Discretization based on the closed sub-ideal {0,3,6} (Lean 4 Mathlib ZMod 9), reducing FLOPs by 68.4%. Evaluated on JevBench (N=231), werr achieves 100.00% TypeSafe compliance and 81.65% calibrated accuracy with 7.08 ms median latency. We provide an OpenAI-compatible API and demonstrate deployment on 32-byte EVM smart contracts via the open-source werracle on-chain oracle (21,438 gas).

Volkan Dağlı, Zerrin Dağlı, Dağhan Dağlı

Anadolu University, Eski¸sehir, Turkey · ITouch Systems, Mersin, Turkey · Mersin University, Mersin, Turkey +1

On-Device Language Model Inference Efficient Language Model Inference Language Model Robustness

Sep 21, 2026cs.LG

Extending FunctionGemma for Practical On-Device Mobile Function Calling

On-device assistants require function-calling models that map natural language to local system actions, but existing resources emphasize web APIs or narrow mobile-action catalogs. We extend FunctionGemma 270M-it to practical Android workflows by introducing MOBILEACTIONSEXTENDED, a synthetic, schema-validated dataset of ~9,500 conversations covering fifteen device-control categories, including messaging, phone calls, camera/screenshot, brightness control, device-status queries, flashlight control, and application management. We fine-tune the 270M model with TRL supervised fine-tuning under completion-only loss, producing an extended specialist and a combined model trained jointly with Google's MOBILEACTIONSGOOGLE. On MOBILEACTIONSEXTENDED, end-to-end accuracy improves from 29.3% for the base model and 17.2% for Google's Mobile-Actions variant to 76.5%. The combined model retains 76.5% on MOBILEACTIONSEXTENDED and reaches 82.3% on MOBILEACTIONSGOOGLE, down from the 90.3% of Google's Mobile-Actions specialist, representing an 8.0-percentage-point trade-off in return for doubling category coverage. We release the dataset, fine-tuned models, reproducible training/evaluation pipeline, and an Android demo, highlighting compact local function calling as a practical path towards low-latency and privacy-preserving mobile assistants.

Ali Rezagholizadeh, Soheila Samiee

UGrowAI

On-Device Language Model Inference Tool-Using Language Agents

Sep 21, 2026cs.LG

LoRA-generating hypernetworks for efficient on-device LLM generative personalization

On-device large language models (

LLMs'), e.g. running on mobile phones, are ripe for improvement via personalization. The limited compute resources of mobile devices impose limits on model scale and thus model quality, making any realizable quality gains highly impactful. At the same time, their personal nature (i.e., the close coupling to a particular user) means that a given on-device LLM tends to be used in similar, predictable patterns over the course of time. This paper presents a novel method for personalizing on-device LLMs. It trains a hypernetwork to map a user's context tokens to a low-rank adaptation (

LoRA') well-suited to that user. Once the trained common artifacts are deployed to users' devices, each user uses the hypernetwork to synthesize (entirely on device) a personalized LoRA. This approach blends the benefits while avoiding the drawbacks of two existing approaches to LLM customization: in-context learning (ICL') and parameter-efficient fine-tuning (PEFT'). Like ICL (and unlike PEFT), the on-device phase of our approach is computationally feasible, requiring only forward passes through neural networks. Like PEFT (and unlike ICL), our approach modifies the `target' base LLM via weights (the LoRA), avoiding negative consequences (e.g. increased latency) associated with extending the input sequence. Our approach is particularly well-suited to the mobile device regime. Apart from the on-device compute and latency benefits mentioned, it also requires minimal additional storage, as internally its architecture partly leverages the same LLM weights as belong to the target LLM to be personalized. We demonstrate the benefits of LoRA-generating hypernetworks on several representative personalization datasets, comparing against baselines like ICL and PEFT. Of note, our personalization experiments focus on more challenging and less studied long-form text generation tasks.

Sean Augenstein, Li Ding, Jihwan Lee +2

Google · Formerly at Google, work done while at Google

On-Device Language Model Inference Low-Rank Adaptation Hypernetworks Personalized Language Models Parameter-Efficient Fine-Tuning

Sep 21, 2026cs.CL

Efficient LLM Distillation for Bangladesh Legal Context: A Smartphone-Compatible Retrieval-Augmented Generation Model

Legal information in Bangladesh is inaccessible to most citizens. Statutory text is English-only, trained lawyers are concentrated in urban centres, and cloud-dependent AI fails where mobile connectivity is unreliable, a setting in which hallucinated legal text causes direct harm. The system addresses statutory interpretation only; queries that require judicial precedent or case-law reasoning fall outside its scope. We target the statutory access gap by compressing a 9-billion-parameter Gemma-2 teacher into a 2-billion-parameter student through two-phase progressive knowledge distillation. Phase 1 performs supervised fine-tuning on 9,429 quality-gated legal question-answer pairs (65% acceptance from 14,514 generated queries); Phase 2 minimises sparse Kullback-Leibler divergence against the teacher's top-50 per-token logits at temperature tau = 4.0, implemented via QLoRA (4-bit NF4, rank-32 LoRA adapters). Prior legal language models target general legal English; this system specialises in Bangladeshi statutory law. Every response is grounded through hybrid retrieval combining dense semantic search (60%) and BM25 (40%) across 36,029 statutory passages from the Bangladesh Constitution and national legislation. On a 50-query English benchmark, the distilled model reaches ROUGE-L 0.4715 and BERTScore F1 0.5679, a 103% ROUGE-L and 143% BERTScore gain over the retrieval-augmented undistilled baseline (ROUGE-L 0.2323, BERTScore 0.2340). The adapter quantises to 1.6 GB (GGUF Q4_K_M) and runs at 4-8 tokens per second on a Pixel 6 with no network access. Cross-lingual evaluation on 50 Bangla queries yields ROUGE-L 0.4083 and BERTScore 0.8133, showing effective retrieval from Bangla input against an English-only corpus. In a single-evaluator pilot, a practising lawyer rated 50 responses at a weighted mean of 4.16/5 (90% rated 4 or 5), supporting utility beyond text-overlap metrics.

MD. Nafis Kamal, Mahadi Hasan Fahim, Talha Ridwan +4

Computer Science and Engineering BRAC University Dhaka, Bangladesh

Retrieval-Augmented Generation Legal NLP On-Device Language Model Inference Language Model Distillation Legal QA

Sep 20, 2026cs.CL

this-that-model-1.0: A typed decision model that decides in 30 ms, for a millionth of a cent

Software delegates more of its branches to models every year: which queue a ticket enters, whether a command is safe to run, whether a claim clears without a person. What the program needs back is not prose. It is one of n declared options and a number it can threshold. Today that costs a round trip to a frontier model -- hundreds of milliseconds, a per-token bill, and a parser -- for a question that is usually a conjunction of three clauses. this-that-model-1.0 is a 2B-parameter typed decision model. Its answer is read directly from the hidden state at a designated position and restricted to the option set the caller declared, so no text is generated, nothing can be malformed, and every question in a request is answered in the same forward pass. It decides in 30.9 ms on one laptop GPU and generates zero output tokens doing it, where a frontier API call costs 8758 ms and the hosted systems that answer these questions well spend between 21 and 212 generated tokens per question thinking first, billed for every one. It sustains 32 decisions per second on one consumer GPU and never lets the state leave the machine. On a third party's recorded cohort of 68 decision questions, on their inputs and their wording, it scores 0.941 with a Brier score of 0.042, against 0.765 and 0.133 for the hosted service Jev on the same items. One pass of our 42-family internal suite takes 32 seconds and 0.000217 USD of electricity; the most accurate hosted model we measured needs 155.2 minutes and 10.636 USD. We also report where it loses. On multi-step arithmetic, which a single forward pass cannot carry intermediate results through, it scores 0.560 against their 0.98 to 1.00, and a targeted second training round improved the five task families it was written for and transferred to none of the other 13. The model is open-sourced in https://huggingface.co/flock-io/this-that-model-1.0

Zehua Cheng, Wei Dai, Jiahao Sun

University of Oxford

On-Device Language Model Inference LLM Decision-Making Efficient Language Model Inference

Sep 20, 2026cs.CL

ARID: A Deployable Edge AI System for Structured Information Extraction from Industrial Maintenance Work Orders

Maintenance work orders must often be processed offline on embedded hardware, yet downstream software requires predictable structured output. We present ARID (Aviation-inspired Routing for Industrial Deployment), which extracts component, failure mode, symptom, and maintenance action into fixed-schema JSON on an 8 GB NVIDIA Jetson Orin NX. ARID combines conservative dual-teacher filtering, targeted noise-aware synthesis, one routing decision per work order, 4-bit inference, and grammar-constrained decoding. From 2,326 unlabeled OMIn records, it retains 716 training pairs and adds 99 topology-constrained records targeting action extraction. On 300 human-labeled records, ARID reaches 84.8% token-F1 on the reference stack and 82.9% on the deployed Jetson. Resident serving achieves 5,310/5,656 ms P50/P99 at 12.5 W. On zero-shot MaintNet transfer, semantic F1 falls to 46.4% while parser success remains at least 99.8%, showing that output validity transfers but field semantics do not.

Kuanlin Chen, Chen-Wei Kuo

Independent Researcher Taoyuan, Taiwan · National Tsing Hua University Hsinchu, Taiwan

On-Device Language Model Inference Document Information Extraction LLM Information Extraction

Sep 18, 2026cs.LG

TierKV: Long-Context On-Device LLMs via Predictive Multi-Tier KV Caching

Large language models (LLMs) are moving onto mobile devices for increasingly diverse workloads over text, images, video, and audio. These applications often require long contexts, making the Key-Value (KV) cache a dominant memory bottleneck because it grows linearly with sequence length and is accessed at every decoding step. Prior work reduces KV-cache footprint through low-rank compression, token eviction, or flash offloading, but the resulting reconstruction overhead, irreversible token loss, or I/O stalls can offset the benefit of saving memory. We present TierKV, a mobile LLM inference framework built on Predictive Multi-Tier Cache Optimization (PMCO). Before decoding starts, PMCO predicts future cache demand from prefill hidden states and jointly assigns tokens to exact, low-rank, and flash-offloaded tiers under the device memory and accuracy budgets. This formulation retains access to the full context, removes the circular dependency of reactive eviction, and admits a closed-form solver that selects tier boundaries and per-layer ranks at runtime. Across eight text, vision, and audio models on three mobile SoCs, TierKV improves prefill throughput by up to 17.6x over existing mobile LLM frameworks, reduces RAM-resident KV cache by 12.5-34%, thereby enabling substantially longer contexts under the same memory budget, while incurring only minor accuracy degradation.

Zhihao Shu, Md Musfiqur Rahman Sanim, Jie Hu +4

University of Georgia Athens, GA, USA · Peking University Beijing, China · Western Digital Research San Jose, CA, USA

On-Device Language Model Inference Long-Context Language Model Inference KV-Cache Management Memory-Efficient Inference

Sep 16, 2026cs.CL

Selection Is Retrieval, Abstention Is Not: On-Device Tool Routing over 70 Korean-English Actions

An AI assistant that calls tools makes two decisions on every request: which tool to invoke, and whether any available tool applies. In the usual design a single language model makes both, by emitting a call or by declining to emit one. On a device that has to answer without a server, the language model is what makes that design expensive, dominating both the latency and the memory of the router. The common alternative is to remove the model completely and rank the catalog of local actions with a retriever instead. That substitution is not symmetric across the two decisions. A retriever returns its highest-scoring candidate for every input and cannot signal that the catalog holds no valid action. Our earlier study found that constraining a decoder to a tool grammar repairs malformed output without improving the choice. What the substitution costs in each decision has not been measured. We evaluate the two decisions separately over 600 Korean and English requests and a catalog of 70 local actions. The router may also ask for a missing slot, reply, or delegate. Half the in-catalog requests reuse catalog vocabulary and half paraphrase it, separating lexical overlap from the action requested. Character 3-gram BM25 selects 162 of 164 lexically matched requests and 85 of 166 paraphrases. Restricting the candidate set to seven raises the paraphrase figure to a mean of 0.825 over five trials. No classifier over its score features separates in-catalog from out-of-catalog above 0.697 area under the curve, where the frozen encoder multilingual-e5-base reaches 0.806. Using that encoder for abstention alone keeps 376 of the requests local and misroutes 9 of the 150 needing delegation. Abstention, not selection, is where a neural component is required. A neural ranker improves every quality metric and is rejected on latency and memory rather than accuracy.

Janghoon Lee

Redrob

Tool-Augmented Language Model Agents On-Device Language Model Inference LLM Abstention Tool Use LLM Agent Routing

Sep 15, 2026cs.AI

JustFit: Just-in-Time State Management for Local LLM Serving

Local agents need memory for model execution and working history. We present JustFit, an MLX runtime that coordinates their overlapping allocations: KVExec executes and checkpoints four-bit KV with bounded workspace, PhaseSwap loads phase-dependent components, and StateTrans preserves history across execution modes. With Qwen3.8-27B MXFP4 on a 24 GiB M4 Pro MacBook, JustFit reaches 327,680 retained positions across two requests, 10.67 times the evaluated baseline's 30,720-position single-request record. One request completes the full 262,144-position native window at median 5.986 tokens/s. Each shape completes 16,384 outputs per request in three fresh processes: B1 uses one cold build and two prefix extensions; B2 uses three ordered prefix extensions (Section 4). Image encoding can proceed while preserving a live 196,608-input text request. A controlled, repetitive 32K+6K workload reaches median 18.284 tokens/s at 15,626 MiB; a separate AIME 2026 evaluation scores 29/30. Coordinating execution and state lifetimes makes longer histories feasible on personal hardware.

Yuhua Chen

LLM Serving On-Device Language Model Inference KV Caching KV-Cache Management Memory-Efficient Inference LLM Agent Serving+1

Sep 14, 2026cs.PF

The Battery Price of edge AI: A study of the Environmental Impact of LLM Inference on Mobile Devices

The rapid diffusion of generative artificial intelligence raises privacy, latency, and performance concerns that motivate a shift toward "local-first" AI, where inferences are performed on the user's device instead of on remote cloud servers. This paradigm also places a significant computational load on battery-powered smartphones, potentially shortening battery life and increasing the overall replacement rate of mobile devices. This paper presents a systematic study of the energy consumption, performance, and accuracy of on-device large language model (LLM) inference. We evaluate 18 models from different model families, sizes, and quantization levels, on two modern smartphones and on a server, using the respective state-of-the-art for such deployments. We measure the energy per generated token, inter-token latency, model accuracy, and battery-cycle consumption. Our results show that (i) on-device inference is on average 3 times less energy-efficient than batched server inference; (ii) the relationship between quantization bit-width and energy per token is non-monotonic, with energy sweet spots on both tested smartphones; (iii) eight out of 18 model configurations lie on the Pareto front of accuracy and energy-efficiency, allowing practitioners to build battery-aware model routers; and (iv) realistic modeling assumptions do not allow local inference to be less environmentally impacting per token than batched server inference, with 88--90% of that impact attributable to device embodied carbon rather than electricity consumption. These findings challenge the premise that local AI is more sustainable than cloud inference, and motivate the need for context-aware and life-cycle-aware model selection when deploying edge AI on battery-powered mobile platforms.

Édouard Guégain, Tristan Coignion

DRI, GREENSPECTOR · Univ. Bordeaux, CNRS, Bordeaux INP, LaBRI, UMR 5800, Talence, France

On-Device Language Model Inference Energy-Efficient ML LLM Inference Responsible AI

Sep 14, 2026cs.LG

On-Device Language Models for Privacy-Preserving Stress Prediction: A Multimodal Evaluation on Mobile Health

Stress is a pervasive determinant of mental health and a key target for mobile health interventions. On-device language models (ODLMs) offer privacy-preserving inference without cloud dependency, yet their feasibility for health prediction under mobile resource constraints remains underexplored. We evaluate ODLMs for multi-modal stress prediction using zero-shot prompting, measuring predictive accuracy alongside latency and throughput. Our results show that objective sensor features marginally outperform subjective self-reports on average, and that lightweight sub-2B models achieve low latency with predictable resource usage. Our findings highlight both the promise and the practical constraints of ODLMs for mobile mental health.

Ibukunoluwa Soyebo, Alyssa Donawa, Rodrigo Aguilar Barrios +2

Ming Hsieh Department of Electrical & Computer Engineering, University of Southern California · Thomas Lord Department of Computer Science, University of Southern California

On-Device Language Model Inference Privacy-Preserving Language Models Multimodal Model Evaluation

Sep 10, 2026cs.LG

From Fixed Keys to Readable Schemas: Small Language Models for Vehicle Agent Function Calls

In-vehicle assistants must translate natural-language requests into accurate vehicle function calls under strict memory and latency constraints, making small language models (SLMs) attractive for on-device deployment. For such models, a key design choice is how the available function surface is presented. Two approaches are to represent each function with a dedicated Functional Token (FT) or provide function schemas directly in the prompt. FTs enable compact inference but are restricted to functions learned during training, whereas Schema-in-Prompt (SIP) can generalize to unseen functions at the cost of longer prompts and higher inference overhead. We introduce a benchmark of 9,822 single-turn examples spanning 79 vehicle functions derived from Android Automotive, including held-out functions and requests requiring refusal. We compare both approaches under matched fine-tuning across four SLMs from 270M to 1.7B parameters. On functions seen during training, scaling provides limited benefit: the 270M model can match the 1.7B model, while the strongest overall performance occurs at 0.6B. On held-out functions, FT achieves zero accuracy by construction, whereas SIP generalizes and improves substantially with scale. On out-of-scope requests, FT can invoke an unavailable function it was trained to emit, while SIP more reliably refuses based on the functions offered. This flexibility comes with higher memory use and latency. Our theoretical analysis explains how SIP enables generalization and why longer schema contexts increase inference cost. Overall, function-surface representation, rather than model scale alone, determines the capabilities and failure modes of SLM-based vehicle function calling.

Hamed Jafarzadeh Asl, Yuanhao Yu, Vahid Partovi Nia

Huawei Noah’s Ark Lab, Canada · Huawei Noah’s Ark Lab · Polytechnique Montreal, Canada

LLM Inference Efficiency On-Device Language Model Inference Small Language Models LLM Tool Use

Sep 9, 2026cs.LG

PELM: Power Efficient On-Device LLM Inference with Speculative Decoding and Dynamic Voltage Frequency Scaling

Deploying Large Language Models (LLMs) directly on mobile platforms at the edge is gaining traction due to a myriad of benefits, such as increased privacy, personalization, and reduced latency. However, LLMs have heavy computational requirements, which are difficult for resource-constrained mobile and edge platforms to fulfill. In addition to limited compute resources, mobile and edge systems often have a compact form factor and lack physical mechanisms to dissipate heat generated from high processor usage rates (e.g., fans) to prevent throttling and reduced processing power, which LLMs can easily cause. To mitigate these effects, prior works have proposed various power governing strategies, such as dynamic voltage and frequency scaling (DVFS), for reducing power and heat generation for heavy computational tasks on mobile platforms. Recently, DVFS methods tailored for mobile LLMs have also been proposed. However, these methods mostly focus on optimizing hardware parameters and processor frequencies, and they fall short under some thermally constrained scenarios. Drawing from recent advances in machine learning, we identify and take advantage of the key insight that not all tokens require full-depth inference to maintain high-quality generation. Motivated by this, we present PELM, a solution that augments traditional DVFS processor frequency tuning with two additional workload-specific knobs: 1) speculative decoding and 2) variable verification depth to expand the optimization space to multiple dimensions for more power efficient on-device LLM inference. In extensive evaluations across hardware platforms and datasets, PELM demonstrates superior performance compared to state-of-the-art power governing methods, with up to 23.1% speedup and 52.4% reduction in energy consumption, while maintaining comparable task performance. The source code is available at https://github.com/imec-nu/PELM.

Weisi Yang, Stephen Xia

Northwestern University Evanston, Illinois, USA

LLM Inference Efficiency On-Device Language Model Inference LLM Inference Acceleration

Sep 7, 2026cs.CL

Beyond Fluent Generation: A CPU Reliability Benchmark for MCP-Style Tool Calling in Sub-2B Small Language Models for Edge Deployment

Resource constrained single-board computers including Raspberry Pi, NVIDIA Jetson Nano, Arduino UNO Q, Orange Pi, and LattePanda motivate on-device small language model (SLM) agents that reduce cloud dependence, improve data locality, and tolerate intermittent connectivity. Model Context Protocol (MCP)-style tool invocation demands more than fluent generation: an agent must emit machine-readable JSON, select the correct tool, supply all required arguments, and avoid unintended actions. We establish a platform-agnostic CPU baseline by evaluating five open-weight models below two billion parameters Phi-1.5, Pythia-1.4B, TinyLlama-1.1B-Chat, Qwen2.5-0.5B, and Qwen2.5-1.5B on 100 prompts spanning weather retrieval, web search, calculation, email composition, and task creation, under greedy decoding and nucleus sampling. A recovery parser strips Markdown fences, extracts brace-delimited substrings, and scores parseability, tool-name correctness, argument completeness, and value agreement. Under this criterion, Qwen2.5-1.5B achieves 75% (greedy) and 79% (sampling); Qwen2.5-0.5B achieves 72% (greedy) but drops to 32% under sampling. Phi-1.5 scores 0%; Pythia and TinyLlama reach at most 7%. A strict post-hoc audit finds only 5 of 1,000 raw responses directly parseable as JSON, exposing near-total dependence on output recovery. A CPU resource probe shows Qwen2.5-1.5B requires 7,960 MiB and 30.782 s mean latency; Qwen2.5-0.5B uses 3,637 MiB and 10.627 s, revealing a reliability-resource trade-off for edge deployment. These results do not cover the named boards directly or a full MCP implementation. Safe deployment requires schema validation, constrained generation, least-privilege execution, and human escalation for consequential actions.

Abrar Shahriar Qurat-Ul-Ain Mastoi

University of the West of England, Bristol,BS16 1QY United Kingdom

Tool-Augmented Language Model Agents On-Device Language Model Inference Language Model Generation Evaluation Small Language Models LLM Reliability LLM Tool Use+1

Sep 7, 2026cs.AI

Diffusion Language Models for Mobile Edge Agentic AI: Foundations, Applications, and Challenges

Diffusion language models (DLMs) offer a non-autoregressive alternative for mobile edge agentic artificial intelligence (AI) by refining tokens through iterative denoising rather than left-to-right decoding. Compared with autoregressive Transformer-based large language models (LLMs), DLMs can update multiple uncertain tokens in parallel and exploit bidirectional context throughout the generation process, enabling more flexible quality-latency trade-offs beyond fixed sequential decoding. These properties are particularly attractive for edge agents, where partial refinement, early exit, and constraint-guided correction can reduce response delay and communication overhead while improving robustness under noisy, incomplete, or dynamic contexts. This survey reviews DLM foundations and analyzes their suitability for edge settings under latency, memory, energy, bandwidth, privacy, and reliability constraints. We cover resource-efficient architectures, training and inference acceleration, compression, edge/cloud deployment, communication-aware serving, Internet of Things (IoT)/wireless applications, and evaluation of DLM-based agents. We further discuss open issues in long-context state management, split inference, trustworthy execution, multimodal grounding, and reproducible benchmarking. The goal is to connect DLM modeling properties, including bidirectionality, parallel refinement, controllability, and quality-latency elasticity, with system-level requirements of future mobile edge intelligence.

Chenqi Li, Minghui Min, Dusit Niyato +1

School of Information and Control Engineering, China University of Mining and Technology, Xuzhou, 221116, Jiangsu, China · College of Computing and Data Science, Nanyang Technological University, Singapore, 639798, Singapore · School of Engineering, Edith Cowan University, Perth, Western Australia, 6027, Australia

On-Device Language Model Inference Diffusion Language Model Inference Diffusion Language Models Diffusion Model Inference Acceleration

Sep 2, 2026cs.LG

LeanStream: A Speculate-and-Refine Streaming Framework for Efficient on-Device LLM Inference

On-device LLM inference is attractive for privacy and responsiveness, but remains challenging on mobile and embedded devices because model weights far exceed available DRAM. Prior systems exploit activation sparsity and offload weights to SSD or flash storage, but face a fundamental systems trade-off: accurate sparse execution decisions require the latest context, whereas efficient computation-I/O overlap requires early prediction. As a result, existing designs either serialize execution or incur redundant weight fetches, extra computation, and large cache overheads. We present LeanStream, a streaming speculate-and-refine framework for efficient on-device LLM inference. LeanStream progressively refines computation, loading, and cache-retention priorities using partial GPU results, enabling fine-grained overlap between GPU execution and storage I/O. We implement LeanStream on both mobile and embedded platforms. Compared with prior on-device LLM inference systems, LeanStream reduces memory usage by 4.8

\times

to 7.5

\times

at the best throughput achieved by prior work, while further improving token generation throughput by 1.6

\times

to 2.1

\times

.

Renyuan Liu, Yuyang Leng, Kaiyan Liu +7

Richard · George Mason University · Global Technology Applied Research, JPMorganChase +1

LLM Inference Efficiency On-Device Language Model Inference Memory-Efficient Optimization Memory-Efficient Inference Efficient Language Model Inference

Sep 1, 2026cs.CL

How Do Prompt Variations Affect Energy Consumption in On-Device LLMs?

Large language models (LLMs) are increasingly deployed on mobile devices, making energy efficiency a key deployment constraint, yet the energy impact of prompt design remains underexplored. This paper aims to understand how two prompt properties, cognitive load and phrasing pattern, shape the energy behavior of on-device LLM inference. We conduct a broad empirical study covering prompt properties, datasets, models, and devices, with phase-level profiling that separates prefill and decode energy. We find that cognitive load primarily affects the energy cost per token, while phrasing pattern affects energy largely through token usage. Our energy-quality analysis further shows that prompt design reshapes the attainable frontier differently across models, highlighting the need for model-aware prompt design in energy-efficient on-device LLM inference. Code, datasets, and scripts are available at https://amai-gsu.github.io/PromptProperty/.

Wei Hu, Xiaolong Tu, Dawei Chen +3

Georgia State University · Toyota Motor North America

LLM Inference Efficiency On-Device Language Model Inference LLM Prompting Efficient Language Model Inference

Sep 1, 2026cs.OS

mzCache: On-Device LLM Memory Management under Multitasking

On-device mobile Large Language Model (LLM) inference is gaining significant attention. However, mobile devices operate in highly dynamic multitasking environments where users frequently switch between applications. This creates memory pressure, forcing LLM memory (model weights and KV cache) to be evicted by the operating system. When a new inference request arrives, the inference system must restore the evicted memory through slow storage reads or recompute the entire KV cache, severely degrading responsiveness. To address this, we present mzCache, an on-device LLM inference system with specialized memory management for multitasking environments. Under unpredictable memory pressure, mzCache elastically evicts LLM memory and leverages the unified memory of mobile SoCs to enable zero-wait inference on the GPU with concurrent CPU-side restoration. mzCache realizes this through restoration-oriented memory management: LLM memory is partitioned into fine-grained shared buffers to enable partial eviction and restoration with concurrent cross-processor access, while hybrid swap and backward-out eviction policies ensure low-latency restoration from any eviction state. Implemented on llama.cpp and deployed as an Android application, mzCache achieves 2.1-5.5

\times

reduction in Time-to-First-Token compared to storage-backed partial offload and demonstrates its effectiveness in real multitasking scenarios.

Hongseung Yu, Minsung Kim, Jongseok Park +1

Seoul National University · UC Berkeley

On-Device Language Model Inference Memory-Efficient Inference LLM Inference Acceleration

Sep 1, 2026cs.AI

Triple-Bottom-Line Sustainability of Language Models for Edge AI: A Comparison Between SLMs and Quantized LLMs

Edge-AI model selection is commonly driven by one isolated metric - accuracy, latency, memory, energy, or safety, even though a deployable language model must balance all five. Our work focuses on answering the question whether na- tively trained small language models (SLMs) or large language models (LLMs) compressed through post-training quantization offer the more sustainable edge- deployment trade-off. We introduce a reproducible Holistic Sustainability Score (HSS) organized around the triple bottom line: an economic pillar for capability and systems efficiency, an environmental pillar for operational GPU energy and a social pillar for harmful-prompt robustness. Five BF16 SLMs and five LLMs under different quantization approaches - BF16, INT8, NF4 4-bit, GPTQ 4-bit, and GGUF Q4 produce 30 measured configurations. Capability is assessed on five zero-shot benchmarks; efficiency uses latency, throughput, peak VRAM and energy; and safety is approximated by attack success rate on five harmful prompts. Qwen3-30B-A3B/GGUF Q4 ranks first in the combined pool (93.38), followed by Mistral-Small-24B/GGUF Q4 (92.40), while Phi-4-mini/BF16 is the highest- ranked SLM in that pool (89.49). Thus, the hypothesis that native SLMs must be the most sustainable edge choice is not supported universally; optimized quantized LLMs can win overall, while SLMs remain competitive through lower resource demand. Quantization is a systems-level choice rather than a monotonic precision- efficiency trade-off and HSS remains relative to its comparison pool and proxy definitions.

Jainil Dharmil Shah

Elmore Family School of Electrical and Computer Engineering · Purdue University, West Lafayette, Indiana, USA

LLM Quantization LLM Evaluation On-Device Language Model Inference Small Language Models

Aug 22, 2026cs.LG

How Weight Encoding Affects Language Model Placement and Performance on the Apple Neural Engine

Weight compression can alter accelerator placement as well as memory traffic, complicating the interpretation of inference speedups. We investigate this interaction on the Apple Neural Engine through the public Core ML deployment path. Five independently trained language-model checkpoints span two architectures and dense fp16, int8, and ternary weights encoded with two-bit lookup tables. We combine compiler device plans, synchronized memory-controller measurements, and compute-unit exclusion controls for a fixed single-token forward workload. On an M1, the smaller fp16 export executes on the CPU despite permitting ANE execution, whereas its compressed counterparts exhibit ANE activity. The int8 export reduces warm forward latency by a factor of 1.9. The larger fp16 export also uses the ANE, indicating that dense encoding alone does not determine placement. Separate M3 energy measurements support the same direction of change. These results establish encoding-dependent placement in the measured deployment stack and show that compression comparisons require joint measurement of backend selection, latency, and traffic.

Shahir M A

Independent researcher

LLM Quantization AI Accelerator Inference On-Device Language Model Inference LLM Inference Acceleration