Daily Papers

BERT (Bidirectional Encoder Representations from Transformers) has revolutionized the field of natural language processing through its exceptional performance on numerous tasks. Yet, the majority of researchers have mainly concentrated on enhancements related to the model structure, such as relative position embedding and more efficient attention mechanisms. Others have delved into pretraining tricks associated with Masked Language Modeling, including whole word masking. DeBERTa introduced an enhanced decoder adapted for BERT's encoder model for pretraining, proving to be highly effective. We argue that the design and research around enhanced masked language modeling decoders have been underappreciated. In this paper, we propose several designs of enhanced decoders and introduce DrBERT (Decoder-refined BERT), a novel method for modeling training. Typically, a pretrained BERT model is fine-tuned for specific Natural Language Understanding (NLU) tasks. In our approach, we utilize the original BERT model as the encoder, making only changes to the decoder without altering the encoder. This approach does not necessitate extensive modifications to the model's architecture and can be seamlessly integrated into existing fine-tuning pipelines and services, offering an efficient and effective enhancement strategy. Compared to other methods, while we also incur a moderate training cost for the decoder during the pretraining process, our approach does not introduce additional training costs during the fine-tuning phase. We test multiple enhanced decoder structures after pretraining and evaluate their performance on the GLUE benchmark. Our results demonstrate that DrBERT, having only undergone subtle refinements to the model structure during pretraining, significantly enhances model performance without escalating the inference time and serving budget.

We show that verifier-free evolution is bottlenecked by both diversity and efficiency: without external correction, repeated evolution accelerates collapse toward narrow modes, while the uniform use of a high-cost model wastes compute and quickly becomes economically impractical. We introduce Squeeze Evolve, a unified multi-model orchestration framework for verifier-free evolutionary inference. Our approach is guided by a simple principle: allocate model capability where it has the highest marginal utility. Stronger models are reserved for high-impact stages, while cheaper models handle the other stages at much lower costs. This principle addresses diversity and cost-efficiency jointly while remaining lightweight. Squeeze Evolve naturally supports open-source, closed-source, and mixed-model deployments. Across AIME 2025, HMMT 2025, LiveCodeBench V6, GPQA-Diamond, ARC-AGI-V2, and multimodal vision benchmarks, such as MMMU-Pro and BabyVision, Squeeze Evolve consistently improves the cost-capability frontier over single-model evolution and achieves new state-of-the-art results on several tasks. Empirically, Squeeze Evolve reduces API cost by up to sim3times and increases fixed-budget serving throughput by up to sim10times. Moreover, on discovery tasks, Squeeze Evolve is the first verifier-free evolutionary method to match, and in some cases exceed, the performance of verifier-based evolutionary methods.

Mixture-of-Experts (MoE) models typically fix the number of activated experts k at both training and inference. However, real-world deployments often face heterogeneous hardware, fluctuating workloads, and diverse quality-latency requirements, while training separate models for each scenario is costly. Considering that MoE models already operate with sparse activation, adjusting the number of activated experts offers a natural path to serving diverse budgets with a single model. Yet, we find that activating more experts k' (> k) at inference does not yield the expected gains. Instead, performance degrades rapidly after only a slight increase, a phenomenon we term the inference-time scaling wall. Further investigation reveals that this degradation stems from a lack of learned collaboration among experts. To address this, we introduce Elastic Mixture-of-Experts (EMoE), a novel training framework that enables MoE models to elastically vary the number of activated experts at inference. By simultaneously training experts to collaborate in diverse combinations and encouraging the router to make high-quality selections, EMoE ensures robust performance across inference budgets. Extensive experiments across four MoE architectures (7B--21B) and nine benchmarks show that EMoE significantly expands the effective scaling range to 2-3times the training-time k, while also achieving higher peak performance.

Mixture of experts (MoE) is a popular technique to improve capacity of Large Language Models (LLMs) with conditionally-activated parallel experts. However, serving MoE models on memory-constrained devices is challenging due to the large parameter size. Typical solutions such as memory swapping or expert pruning may lead to significantly higher latency or severe accuracy loss. In this paper, we introduce SwapMoE, a framework for efficient serving of MoE-based large language models with tunable memory budgets. The main idea of SwapMoE is to keep a small dynamic set of important experts, namely Virtual Experts, in the main memory for inference, while seamlessly maintaining how the Virtual Experts map to the actual experts. Experiments have shown that SwapMoE can reduce the memory footprint while maintaining reasonable accuracy. For example, on text summarization tasks with Switch Transformer, SwapMoE can reduce the memory consumption from 14.2 GiB to 4.7 GiB, together with 50\% latency reduction and a slight Rouge-2 score drop of 0.041.

Efficient deployment of large language models (LLMs) in production forces a trade-off between accuracy and cost. Operators often default to a single model that is either expensive for easy queries or insufficient for hard ones. To address this challenge, we propose a two-stage cascaded solution. Stage 1 clusters incoming queries and assigns each cluster to its most cost-effective model. The cost budget for this routing process is set by an interpretable hyperparameter, tuned offline. Stage 2 adds a quality estimation (QE) cascade; when an output from Stage 1 is judged low-quality, the query is escalated to a stronger model. This ensures only hard or low-confidence cases reach the expensive models. On the test datasets, the cascaded system retains 97-99% of the strongest model's accuracy while reducing Time Per Output Token (TPOT). It requires only task-correctness labels and adapts to changes in the model pool without manual reconfiguration.

Diffusion Large Language Models (dLLMs) have emerged as a promising alternative to Autoregressive Models (ARMs), utilizing parallel decoding to overcome sequential bottlenecks. However, existing research focuses primarily on kernel-level optimizations, lacking a holistic serving framework that addresses the unique memory dynamics of diffusion processes in production. We identify a critical "memory footprint crisis" specific to dLLMs, driven by monolithic logit tensors and the severe resource oscillation between compute-bound "Refresh" phases and bandwidth-bound "Reuse" phases. To bridge this gap, we present dLLM-Serve, an efficient dLLM serving system that co-optimizes memory footprint, computational scheduling, and generation quality. dLLM-Serve introduces Logit-Aware Activation Budgeting to decompose transient tensor peaks, a Phase-Multiplexed Scheduler to interleave heterogeneous request phases, and Head-Centric Sparse Attention to decouple logical sparsity from physical storage. We evaluate dLLM-Serve on diverse workloads (LiveBench, Burst, OSC) and GPUs (RTX 4090, L40S). Relative to the state-of-the-art baseline, dLLM-Serve improves throughput by 1.61times-1.81times on the consumer-grade RTX 4090 and 1.60times-1.74times on the server-grade NVIDIA L40S, while reducing tail latency by nearly 4times under heavy contention. dLLM-Serve establishes the first blueprint for scalable dLLM inference, converting theoretical algorithmic sparsity into tangible wall-clock acceleration across heterogeneous hardware.

Streaming video generation is emerging as a new serving workload in which users interact with long-lived sessions that generate video progressively, chunk by chunk. Unlike offline video generation or typical LLM serving, streaming video generation must preserve session state across active and idle periods, repeatedly schedule ongoing sessions, and deliver each chunk under a tight latency target. This creates two key serving challenges in multi-user, multi-GPU environments: session duration heterogeneity, where long-running sessions make placement decisions suboptimal over time, and temporal user-demand heterogeneity, where the number of active sessions fluctuates sharply across bursts and idle periods. We present TurboServe, the first serving system designed specifically for streaming video generation workloads. TurboServe formulates serving as an online scheduling problem that jointly coordinates session placement and GPU provisioning. Its closed-loop scheduling algorithm combines a migration-aware placement controller, which rebalances sessions across GPUs to reduce the maximum per-chunk latency, with a load-driven autoscaling controller, which adapts the GPU budget to workload variation for improved cost efficiency. To support these decisions at runtime, TurboServe implements coalesced chunk processing for batching concurrent active sessions on the same GPU, GPU-CPU offloading for session suspension and resumption, and NCCL-based GPU-GPU migration for online rebalancing. We evaluate TurboServe on real-world production traces from Shengshu Technology across multiple model sizes and GPU clusters with up to 64 NVIDIA B300 GPUs. Compared with baseline serving configurations, TurboServe reduces worst-case per-chunk latency by 37.5% and total GPU operating cost by 37.2% on average. Our code is publicly available at https://github.com/shengshu-ai/TurboServe.

We study linear dueling bandits in volatile environments characterized by the simultaneous presence of post-serving contexts, delayed feedback, and adversarial corruption. Feedback is subject to unknown stochastic or adversarial delays and a cumulative corruption budget C. To address these challenges, we propose e RCDP-UCB, which integrates a learned approximator that predicts post-serving contexts from pre-serving information. It further employs an adaptive weighting strategy that clips feature vectors to mitigate the impact of corrupted and delayed observations simultaneously. Under standard regularity conditions and a parametric post-serving mapping, we rigorously establish that our algorithm is delay-regime-agnostic, achieving a regret upper bound of mathcal{O}(d(T + C + D)), where d is the total feature dimension and D encapsulates the delay complexity. Crucially, our analysis reveals an additive cost structure between corruption and delay, avoiding the multiplicative degradation typical of prior works. We further establish lower bounds that nearly match our upper bounds up to a d factor for adversarial delays in the absence of post-serving contexts. Code is available at https://github.com/youngmin0oh/rcdp-public.

Test-time reasoning is increasingly used as a serving-time control knob, but extra reasoning is not uniformly valuable: it can repair failed attempts, waste compute on already-correct answers, or introduce harmful answer changes. We study this as a deployment allocation problem rather than a new-verifier problem. We introduce \sevra, Selective Verification for Reasoning Allocation, a serving-layer controller that decides whether to preserve a frozen solver's initial answer or invoke active verification. Using a frozen Qwen3-4B solver, we log intervention outcomes and train recoverability-aware gates from serving-visible attempt state. On \mathfive, selective verification reaches 76.3\% accuracy, compared with 75.5\% for always verifying, while reducing post-generation tokens by 26.8\% and harmful flips from 2.2\% to 1.0\%. However, an 8,192-token initial solve reaches 76.0\% accuracy with 28\% fewer total model tokens, showing that selective recovery is useful but not the best tested cost frontier. In frozen transfer to \gsm, the selective policy verifies only 3.0\% of examples, improves accuracy from 93.4\% to 94.5\%, and reduces verification tokens by 91.2\% relative to always verifying; again, a longer initial solve matches its accuracy with fewer realized tokens. On CommonsenseQA, always-on verification hurts, while Self-Consistency@5 improves accuracy at about five times the realized token cost. The resulting deployment rule is: tune the initial budget first, then use selective recovery when explicit checks, bounded retries, auditability, or regression-risk control matter.

Multi-turn LLM serving accumulates dialogue history whose Key-Value (KV) cache grows with every turn and every user, quickly exceeding the model weights themselves and making memory -- not compute -- the binding constraint on throughput. Non-uniform KV compression, which allocates heterogeneous budgets across attention heads, preserves accuracy far better than uniform schemes, yet remains impractical: modern serving stacks assume identical KV lengths across heads, so heterogeneity traps freed memory as page fragmentation, spends up to 25% of prefill time reclaiming scattered pages, and skews GPU workloads that inflate decode latency by up to 1.7times or burn 15--20% of each decode step on re-planning. We observe that this heterogeneity need not be discovered at runtime: head-wise retention follows a two-level structural regularity -- an input-invariant head ranking with narrowly bounded per-head ratios -- that can be calibrated offline from as few as 50 samples. Building on this insight, we present Tangram, a serving framework that statically resolves what prior systems handle dynamically: Budget Reservation fixes each head's post-compression footprint at scheduling time, eliminating page reclamation; Ragged Paging clusters similar-budget heads into independent page tables, turning fragmentation into reclaimable memory; and Ahead-of-Time Load Balancing precomputes balanced GPU partitions with zero runtime planning. Implemented on vLLM, Tangram serves as a drop-in substrate for existing non-uniform compression methods, matching their accuracy while improving end-to-end throughput by up to 2.6times over the full-KV baseline. Our implementation is publicly available at https://github.com/aiha-lab/TANGRAM.

LLMs are widely adopted in production, pushing inference systems to their limits. Disaggregated LLM serving (e.g., PD separation and KV state disaggregation) improves scalability and cost efficiency, but it also turns KV into an explicit payload crossing network and storage boundaries, making KV a dominant end-to-end bottleneck. Existing KV compression are typically static runtime configurations, despite production service context varies over time in workload mix, bandwidth, and SLO/quality budgets. As a result, a fixed choice can be suboptimal or even increase latency. We present KVServe, the first service-aware and adaptive KV communication compression framework for disaggregated LLM serving: KVServe (1) unifies KV compression into a modular strategy space with new components and cross-method recomposition; (2) introduces Bayesian Profiling Engine that efficiently searches this space and distills a 3D Pareto candidate set, reducing 50times offline search overhead; and (3) deploys a Service-Aware Online Controller that combines an analytical latency model with a lightweight bandit to select profiles under constraints and correct offline-to-online mismatch. Integrated into vLLM and evaluated across datasets, models, GPUs and networks, KVServe achieves up to 9.13times JCT speedup in PD-separated serving and up to 32.8times TTFT reduction in KV-disaggregated serving.

Large Language Model (LLM) inference in production must meet stringent service-level objectives for both time-to-first-token (TTFT) and time-between-token (TBT) while maximizing throughput under fixed compute, memory, and interconnect budgets. Modern serving systems adopt stall-free scheduling techniques such as chunked prefill, which splits long prompt processing along the token dimension and interleaves prefill with ongoing decode iterations. While effective at stabilizing TBT, chunked prefill incurs substantial overhead in Mixture-of-Experts (MoE) models: redundant expert weight loads increase memory traffic by up to 39% and inflate energy consumption. We propose layered prefill, a new scheduling paradigm that treats transformer layer groups as the primary scheduling unit. By vertically partitioning the model into contiguous layer groups and interleaving prefill and decode across the groups, layered prefill sustains stall-free decoding while eliminating chunk-induced MoE weight reloads. It reduces off-chip bandwidth demand, lowering TTFT by up to 70%, End-to-End latency by 41% and per-token energy by up to 22%. Evaluations show that layered prefill consistently improves the TTFT--TBT Pareto frontier over chunked prefill, reducing expert-load traffic and energy cost while maintaining stall-free decoding. Overall, shifting the scheduling axis from tokens to layers unlocks a new operating regime for high-efficiency, energy-aware LLM serving in co-located environments.

Increasing demand for Large Language Models (LLMs) services imposes substantial deployment and computation costs on providers. LLM routing offers a cost-efficient solution by directing queries to the optimal LLM based on model and query features. However, existing works primarily focus on offline scenarios and struggle to adapt to online settings with high query volume and constrained token budgets. In this work, we introduce the first training-free algorithm for online routing scenarios. Our algorithm leverages approximate nearest neighbor search to efficiently estimate query features and performs a one-time optimization over a small set of initial queries to learn a routing strategy that guides future routing. We provide theoretical guarantees demonstrating that our algorithm achieves a competitive ratio of 1 - o(1) under natural assumptions, which is further validated by extensive experiments across 3 benchmark datasets and 8 baselines, showing an average improvement of 3.55times in overall performance, 1.85times in cost efficiency, and nearly 4.25times in throughput. Our code is available at https://github.com/fzwark/PORT.

Entity Linking (EL) and Relation Extraction (RE) are fundamental tasks in Natural Language Processing, serving as critical components in a wide range of applications. In this paper, we propose ReLiK, a Retriever-Reader architecture for both EL and RE, where, given an input text, the Retriever module undertakes the identification of candidate entities or relations that could potentially appear within the text. Subsequently, the Reader module is tasked to discern the pertinent retrieved entities or relations and establish their alignment with the corresponding textual spans. Notably, we put forward an innovative input representation that incorporates the candidate entities or relations alongside the text, making it possible to link entities or extract relations in a single forward pass and to fully leverage pre-trained language models contextualization capabilities, in contrast with previous Retriever-Reader-based methods, which require a forward pass for each candidate. Our formulation of EL and RE achieves state-of-the-art performance in both in-domain and out-of-domain benchmarks while using academic budget training and with up to 40x inference speed compared to competitors. Finally, we show how our architecture can be used seamlessly for Information Extraction (cIE), i.e. EL + RE, and setting a new state of the art by employing a shared Reader that simultaneously extracts entities and relations.

The mixture of experts (MoE) model is a sparse variant of large language models (LLMs), designed to hold a better balance between intelligent capability and computational overhead. Despite its benefits, MoE is still too expensive to deploy on resource-constrained edge devices, especially with the demands of on-device inference services. Recent research efforts often apply model compression techniques, such as quantization, pruning and merging, to restrict MoE complexity. Unfortunately, due to their predefined static model optimization strategies, they cannot always achieve the desired quality-overhead trade-off when handling multiple requests, finally degrading the on-device quality of service. These limitations motivate us to propose the D^2MoE, an algorithm-system co-design framework that matches diverse task requirements by dynamically allocating the most proper bit-width to each expert. Specifically, inspired by the nested structure of matryoshka dolls, we propose the matryoshka weight quantization (MWQ) to progressively compress expert weights in a bit-nested manner and reduce the required runtime memory. On top of it, we further optimize the I/O-computation pipeline and design a heuristic scheduling algorithm following our hottest-expert-bit-first (HEBF) principle, which maximizes the expert parallelism between I/O and computation queue under constrained memory budgets, thus significantly reducing the idle temporal bubbles waiting for the experts to load. Evaluations on real edge devices show that D^2MoE improves the overall inference throughput by up to 1.39times and reduces the peak memory footprint by up to 53% over the latest on-device inference frameworks, while still preserving comparable serving accuracy as its INT8 counterparts.

Mainstream LLM serving systems reuse prefix work mainly through paged or radix key-value (KV) caches. This is highly effective for high-throughput, high-concurrency serving, but it manages only one positional fragment of execution state: the KV cache. We study the opposite regime: low-latency, small-batch, on-device physical-AI serving, where interactive LLM agents, speech systems, and robot policies repeatedly branch, reset, interrupt, and re-enter under tight responsiveness budgets. We introduce execution-state capsules, a graph-bound checkpoint and restore mechanism for the complete restorable state at a committed boundary. FlashRT is a white-box, backend-facing kernel runtime whose evaluated NVIDIA CUDA backend runs captured graph plans over contiguous static buffers with no block-table indirection. Because the live state is a closed set of named buffers, a capsule can snapshot, restore, fork, or roll back the whole execution boundary, including KV, recurrent state, convolution state, MTP state, and metadata. This moves reuse from token-addressed KV fragments to graph-bound execution-state boundaries. On an RTX 5090, capsule restore is byte-exact at the stored-state level and token-identical under greedy decode. A KV-only ablation diverges, showing that recurrent state is load-bearing. GPU-resident snapshot and restore are sub-millisecond, and TTFT speedup over cold prefill grows from 3.9x at 2k tokens to 27x at 16k tokens. On Jetson AGX Thor and DGX Spark, the same correctness and structural properties hold. Capsules are not a replacement for high-throughput KV-cache serving; they define a complementary latency-first serving point for explicit execution-state reuse.

The current pretraining paradigm for large language models relies on massive compute and internet-scale raw text, creating a significant barrier to foundational research. In contrast, biological systems demonstrate highly sample-efficient learning through multi-timescale processing, such as the functional organization of the frontoparietal loop. Taking this as inspiration, we introduce HRM-Text, which replaces standard Transformers with a Hierarchical Recurrent Model (HRM) that decouples computation into slow-evolving strategic and fast-evolving execution layers. To stabilize this deep recurrence for language modeling, we introduce MagicNorm and warmup deep credit assignment. Furthermore, instead of standard raw-text pretraining, we train exclusively on instruction-response pairs using a task-completion objective and PrefixLM masking. Serving as an empirical existence proof of efficient pretraining, a 1B-parameter HRM-Text model trained from scratch on only 40 billion unique tokens and $1,500 budget achieves 60.7% on MMLU, 81.9% on ARC-C, 82.2% on DROP, 84.5% on GSM8K, and 56.2% on MATH. Despite utilizing roughly 100-900x fewer training tokens and 96-432x less estimated compute than standard baselines, HRM-Text performs competitively with 2-7B parameter open models. These results demonstrate that co-designing architectures and objectives can radically reduce the compute-to-performance ratio, making pretraining from scratch accessible to the broader research community.

Mixture-of-Experts (MoE) language models scale model ability with sparsely activated experts, making this architecture a standard recipe for modern large models. However, sparse activation does not remove the deployment burden of storing and serving all experts, and the available deployment budget can vary substantially across devices, users, and workloads. Existing MoE compression methods are still largely fixed-budget, typically optimizing one compressed endpoint at each chosen target budget. We study a different setting: converting a large pretrained MoE LLM into a nested family of deployable subnetworks across budgets. Our method first ranks expert FFN channels by their importance, then lets each expert learn a discrete action to prune its channels. By gradually increasing cost pressure, a single action-training run exports a series of action masks from high to low budgets, each of which identifies a reliable smaller subnetwork nested in the ranked base model. Moreover, we use a single recovery fine-tune at a mid pruning budget (40%) to recover degraded model quality and transfer the recovered model to other unseen budgets. Overall, our framework surpasses recent MoE compression baselines. Specifically, on Qwen2-57B-A14B, our method retains ~99.8% of base performance while pruning 50% of routed expert parameters even without fine-tuning. For deployment, our pruned subnetworks deliver real memory reduction and throughput gains, and further support realtime online budget switching with kernel-level co-design.

Test-time scaling (TTS) boosts LLM reasoning via long chain-of-thought (CoT), but the linear KV-cache growth amplifies the memory-bound bottleneck of LLM decoding. Query-aware page-level sparse decoding can achieve state-of-the-art performance under constrained FLOPs budgets, but is limited by both sequential-dependent page filtering and coarse-grained token selection, hampering serving efficiency and model performance on TTS tasks under high concurrency and long CoT scenarios (consuming even higher runtime than the forward pipeline itself). In this paper, we first find that the current-step query state can be accurately approximated in a unified manner from a short window of recent queries, enabling training-free query-aware sparsity without waiting in the decoding loop. We propose AsyncSpade, an asynchronous framework for efficient TTS built on two core components: (1) a novel light-weight temporal-regressive module that predicts the next-token query state; (2) an asynchronous and disaggregated framework that decouples the KV cache filtering from the auto-regressive decoding loop, overlapping the token-level KV selection with the forward inference computation through asynchronism. To our knowledge, AsyncSpade is the first to eliminate the sequential dependence without sacrificing model performance. We validate the effectiveness of AsyncSpade on common LLM serving setups with an A100 node, where AsyncSpade fully overlaps KV-cache operations with the inference pipeline, achieving theoretical optimal time-per-output-token (TPOT). Specifically, AsyncSpade delivers over 20% reduction on TPOT compared to SoTA baseline (i.e. Quest) and at least 50% TPOT reduction compared to full attention on Qwen3-8B and Qwen3-32B models, while matching or surpassing their accuracy on various TTS benchmarks (AIME-24/25, GPQA-Diamond, MATH-500).

AI agents are increasingly used to automate research and development tasks, yet existing benchmarks typically evaluate them on prescribed workflows or narrow action spaces. Even nominally open-ended tasks can often be solved by retrieving a well-known recipe and tuning a few hyperparameters, making it unclear whether strong results reflect genuine optimization or memorized solutions. We introduce InferenceBench, where an agent must deploy an OpenAI-compatible inference server and optimize the speed of LLM inference. Each agent receives a target LLM, one H100 GPU, an optimization scenario, and a wall-clock time budget of two hours. Three optimization scenarios isolate distinct bottlenecks of inference (prefill latency, decode latency, and concurrent request throughput) and a fourth balances all three at the same time. Across 15 frontier agent configurations, agents reliably improve over a naive PyTorch baseline (up to 8.08times) and often match or exceed serving engines with default settings (4.05times for vLLM), but still fall below a simple hyperparameter search under the same time budget (up to 11.53times). Qualitative analysis of agent trajectories shows that although agents enumerate many relevant optimization techniques, they overwhelmingly converge on a single inference framework. They test only a few distinct configurations and spend the remaining budget re-measuring, repairing, or optimizing hyperparameters rather than exploring substantially different strategies. This suggests the bottleneck is not domain knowledge, but the ability to propose diverse configurations, evaluate them systematically, and submit the best identified solution. Overall, InferenceBench reflects the ability of agents to operate in an open-ended AI engineering setting, where memorized solutions lead to limited improvements.

This position paper argues that agentic AI systems should be designed and evaluated as marginal token allocation economies rather than as text generators priced by the unit. We follow a single request -- a developer asking a coding agent to fix a failing test -- through four economic layers that today are designed in isolation: a router that decides which model answers, an agent that decides whether to plan, act, verify, or defer, a serving stack that decides how to produce each token, and a training pipeline that decides whether the trace is worth learning from. We show that all four layers are solving the same first-order condition -- marginal benefit equals marginal cost plus latency cost plus risk cost -- with different index sets and different prices. The framing is deliberately minimal: we do not propose a complete theory of AI economics. But adopting marginal token allocation as the shared accounting object explains why systems that locally minimize tokens globally misallocate them, predicts a small set of recurring failure modes (over-routing, over-delegation, under-verification, serving congestion, stale rollouts, cache misuse), and points to a concrete research agenda in token-aware evaluation, autonomy pricing, congestion-priced serving, and risk-adjusted RL budgeting.

We survey the large language model (LLM) serving area to understand the intricate dynamics between cost-efficiency and accuracy, which is magnified by the growing need for longer contextual understanding when deploying models at a massive scale. Our findings reveal that works in this space optimize along three distinct but conflicting goals: improving serving context length (C), improving serving accuracy (A), and improving serving performance (P). Drawing inspiration from the CAP theorem in databases, we propose a CAP principle for LLM serving, which suggests that any optimization can improve at most two of these three goals simultaneously. Our survey categorizes existing works within this framework. We find the definition and continuity of user-perceived measurement metrics are crucial in determining whether a goal has been met, akin to prior CAP databases in the wild. We recognize the CAP principle for LLM serving as a guiding principle, rather than a formal theorem, to inform designers of the inherent and dynamic trade-offs in serving models. As serving accuracy and performance have been extensively studied, this survey focuses on works that extend serving context length and address the resulting challenges.

Memory is increasingly central to Large Language Model (LLM) agents operating beyond a single context window, yet most existing systems rely on offline, query-agnostic memory construction that can be inefficient and may discard query-critical information. Although runtime memory utilization is a natural alternative, prior work often incurs substantial overhead and offers limited explicit control over the performance-cost trade-off. In this work, we present BudgetMem, a runtime agent memory framework for explicit, query-aware performance-cost control. BudgetMem structures memory processing as a set of memory modules, each offered in three budget tiers (i.e., Low/Mid/High). A lightweight router performs budget-tier routing across modules to balance task performance and memory construction cost, which is implemented as a compact neural policy trained with reinforcement learning. Using BudgetMem as a unified testbed, we study three complementary strategies for realizing budget tiers: implementation (method complexity), reasoning (inference behavior), and capacity (module model size). Across LoCoMo, LongMemEval, and HotpotQA, BudgetMem surpasses strong baselines when performance is prioritized (i.e., high-budget setting), and delivers better accuracy-cost frontiers under tighter budgets. Moreover, our analysis disentangles the strengths and weaknesses of different tiering strategies, clarifying when each axis delivers the most favorable trade-offs under varying budget regimes.