Inference Cost

Production LLM applications rarely receive a question nobody has asked before. Support assistants and RAG pipelines field the same intents thousands of times a day, each phrased differently, and most stacks treat every phrasing as a fresh, fully billed request. Redis LangCache is a fully managed semantic caching service that sits between the application and […] The post Meet Redis LangCache: A Managed Semantic Cache That Cuts LLM API Costs by Up to 90% and Returns Cache Hits Up to 15x Faster appeared first on MarkTechPost.

Amazon SageMaker Inference now offers prefix-aware routing, a routing strategy that sends requests sharing the same prompt prefix to the same instance so the KV cache stays warm. In benchmarks on Llama 3.1 70B, it reduced P50 time-to-first-token by up to 77% and raised KV cache hit rates from about 25% to over 80%.

NVIDIA has detailed BioNeMo Inference Runtime (BioIR), a Python library that accelerates biomolecular structure-prediction models on NVIDIA GPUs while staying in plain PyTorch. In a matched benchmark on 1,000 human dimer targets across 8xH100 GPUs, BioIR-accelerated Boltz-2 delivered 58.5K successfully folded residues per GPU-hour versus 20.2K for a torch-compiled open-source implementation, a 2.90x gain. The runtime optimizes at 3 layers: custom kernel selection, CUDA Graph capture, and Ray-based replica scaling that places 1 full model copy per GPU. BioIR already powered the AlphaFold Database expansion, generating about 31 million candidate protein complexes across 4,777 proteomes. The post NVIDIA Details BioNeMo Inference Runtime (BioIR): 2.90x Higher Boltz-2 Folding Throughput and 58…

Long-horizon agents have turned LLM serving into an input-heavy workload. Repeated prefills and million-token contexts leave KV caches that strain HBM, SSD capacity, and bandwidth. DeepSeek AI built its newest release around that exact bottleneck. DeepSeek-V4.1-Flash is a multimodal Mixture-of-Experts model with 552B backbone parameters, 196B additional Engram parameters, and a 1M-token context window. It […] The post DeepSeek AI Released DeepSeek-V4.1-Flash with 1M Context, FP4 KV Cache, and Cross-Layer Attention Reuse appeared first on MarkTechPost.

arXiv:2609.09338v1 Announce Type: new Abstract: Speculative decoding is critical for accelerating LLM inference. However, the speedup is fragile: drafters are typically trained against a narrow distribution for a single target model, and their acceptance rate collapses under workload shifts. This is a striking inversion of modern LLM development, where target models are valued precisely for the broad generalization they acquire through large-scale pretraining. We argue that the natural remedy, pretraining, has been hard to apply to drafters because existing recipes are target-specific: the drafter consumes the target's hidden states and is distilled on the target's logits, so pretraining must be repeated for each target. We introduce Osprey, which instead bootstraps drafters from off-the-…

Learn how to deploy Qwen3.8-2.4T-A95B, a 2.4-trillion-parameter open-weight model, on Amazon SageMaker HyperPod with vLLM. This walkthrough covers cluster provisioning, NVFP4 quantization, and an OpenAI-compatible endpoint with built-in reasoning, tool calling, and native MTP speculative decoding.

As LLM applications grow more complex, inference cost and latency become increasingly important. This article breaks down four caching techniques used in LLM serving—KV caching, prefix caching, prompt caching, and semantic caching—and explains how each one avoids repeated computation at a different layer of the stack.

Pretrained vision-language-action (VLA) models handle many manipulation tasks but are not reliable enough for tasks requiring precision and repeatability. VLA-Precision applies real-world online RL to VLA post-training, introducing the Asymmetric Co-Bootstrapping (ACoB) algorithm and the ACoB-Stream architecture. ACoB combines early intervention-guided learning with global return propagation and local preference ranking to suppress policy drift; ACoB-Stream uses invariant-state decoupling and on-demand streaming to improve throughput and computational efficiency by up to 10.9x. Evaluated on nine high-precision chemistry tasks across four robot embodiments, VLA-Precision reaches a 98.3% mean success rate in 45.8 min/task and runs at 1.2x and 1.8x the speeds of VLA and RL baselines.

Perplexity's engineering team published an in-depth look at the GPU serving infrastructure behind pplx-embed. The system reuses LLM prefill/decode kernels, separates responsibilities into a Rust gateway (Ivy), a gRPC server (Tulip), and a Python inference engine (ROSE), and uses CUDA graphs, lazy capture, and LazyTensor to optimize throughput and latency. Benchmarks against vLLM are also covered.

AdaptiveSpec is a training-free, per-step speculative decoding method that adapts both token-acceptance margin and draft-tree shape from internal decoding signals. Implemented on SGLang, it improves throughput over EAGLE-3 by up to 56% while recovering 93% to fully lossless task accuracy on GSM8K, MATH-500, and HumanEval.

A physics-informed graph attention network surrogate runs directly on tetrahedral TCAD meshes and predicts electrostatic potential plus electron and hole quasi-Fermi levels at every mesh node. Combining a data loss with finite-volume current-continuity residuals embeds drift-diffusion physics into training, while graph-based operation lets models trained on few-fin FinFETs transfer to substantially larger multi-fin arrays. Deep-ensemble uncertainty powers active learning; benchmarks against Sentaurus Device show sub-volt field RMSE and orders-of-magnitude higher per-design throughput, with inference still under one second for large arrays.

FORGE introduces a forward-only test-time adaptation method for convolutional vision models deployed on microcontrollers in a quantization-friendly, inference-only setting. Because integer deployment fuses batch-normalization layers into convolutions, it destroys the statistics that normalization-based adaptation needs. FORGE restores adaptation by re-normalizing each folded convolution's per-channel outputs to clean training statistics using only forward-pass estimates. It recovers most of TENT's accuracy gain (+20.9 vs +24.9 points), needs just 3 of 21 layers to capture 93% of the benefit, works with single-sample streaming, and costs only 8.3 mJ / 21.9 ms on an ESP32-S3.

Anthropic has announced Fable 5.1, cutting cache read pricing, and says the model is comparable to Mythos 5.1 but with different safeguards.

Local LLM model names like 'Qwen3.8-27B-A3B-It-2507-gguf-q2ks-mixed-AutoRound' look cryptic but each segment conveys crucial details. This guide breaks down meaning of parameter count, MoE architecture, active parameters, Base/Instruct tuning, weight precision (FP16/BF16), quantization (Q4/Q8), quantization variants (e.g., Q4_K_M), and file format (GGUF) to help you choose the right model.

StreamScout is an adaptive inference framework for streaming video understanding that maintains a lightweight textual timeline and progressively augments it with up to three visual views of increasing detail, only escalating when needed. This reduces inference cost and token consumption while improving accuracy. On OVO-Bench, StreamScout-S improves Qwen3-VL-8B by 14.65 accuracy points while using 59% fewer tokens than uniform sampling and achieving an average response time of 1.04 seconds.

Proposes REAL-Q, a novel post-training quantization (PTQ) method that directly optimizes an end-to-end aligned surrogate loss using fine-grained dynamic block-wise gradient descent and a sliding window mechanism, reducing end-to-end KL divergence by up to ~49% on LLaMA-3.1 and Qwen3 models compared to state-of-the-art methods.

Anthropic has released Claude Fable 5.1 and Claude Mythos 5.1, the same model with different safety layers. Fable 5.1 is generally available, while Mythos 5.1 remains restricted. Fable 5.1 scores 52.6% on Terminal-Bench-Science 0.1, up from Fable 5's 24.7%, and has a 1M token context window. Cache read costs drop 75% to $0.25 per million tokens, with base pricing unchanged. Three breaking API changes affect agents, including the removal of forced tool use.

arXiv:2608.28670v1 Announce Type: new Abstract: Diffusion Transformers achieve high-fidelity image and video generation, but their iterative sampling remains expensive, for each denoising step requires large matrix operations. Existing cache-based acceleration reduces redundant computation yet increases the VRAM footprint by storing intermediate states, which can directly constrain inference batch size. In this work, we propose a training-free acceleration method that performs stepwise forecasting for DiT sampling using a Barycentric Extrapolator. By leveraging barycentric extrapolation, our predictor is numerically stable and alleviates oscillatory artifacts analogous to the Runge phenomenon during forward forecasting. Across extensive experiments on both image and video generation, our…

arXiv:2608.28911v1 Announce Type: new Abstract: The key-value (KV) cache is the dominant memory bottleneck of long-context large language model (LLM) inference, growing linearly with context length. We show that uniform KV quantization on a fractional-bit grid does not degrade gracefully: under a prespecified multi-seed statistical protocol, Llama-3.1-8B-Instruct with an affine quantizer is statistically indistinguishable from FP16 KV down to 2.322 code bits/value and collapses at 2.0 bits - a quality cliff in (2.0, 2.322] that reappears in generation-time quantization and multi-turn dialogue and transfers to Mistral-7B. The cliff reframes importance-aware mixed precision: above it, eight model-internal importance indicators are statistically interchangeable, so the benefit of mixing is g…

arXiv:2608.28599v1 Announce Type: new Abstract: Clinical diagnosis is a step-by-step, cost-aware process: a physician orders examinations one at a time, observes the results, and updates the diagnosis before reaching a final conclusion. Most medical language models instead treat diagnosis as a one-pass classification task and ignore the trade-off between a test's value and its cost. We model diagnosis as a cost-aware sequential decision process and train the policy with reinforcement learning. The main difficulty is credit assignment: the only reliable signal comes once at the end of a long trajectory, so it scores a wasteful workup the same as an efficient one. We propose CDPR (Counterfactual Diagnostic Process Reward), which needs no expert labels and no learned critic. CDPR first finds…

arXiv:2608.28075v1 Announce Type: new Abstract: Manipulation in partially observable environments requires planning under incomplete scene information. In such settings, an initially valid plan may execute successfully yet remain insufficient for task completion. Existing foundation-model-guided task and motion planning (TAMP) systems can generate useful long-horizon task decompositions, subgoals, or constraints, but they often assume having access to a fully specified scene state or invoke model-level replanning after a subgoal, refinement, or execution attempt fails. We present ROBUST TAMP, a modular LLM/VLM-guided planning framework for reactive TAMP where unseen task-relevant and non-target objects may become visible during execution. The framework restricts the foundation-model plann…

arXiv:2608.27685v1 Announce Type: new Abstract: Simultaneously optimizing the trajectories of multiple agents is a challenging problem plagued by nonlinearity, nonconvexity, and the curse of dimensionality. A collection of interacting aerial vehicles or self-driving cars in an intersection are examples of complex multi-agent systems that remain difficult to solve without many simplifying assumptions. The presence of communication latency between agents further increases the difficulty. In this paper, we analyze Distributed Model-Based Diffusion: a sampling-based Model-Predictive Control method suitable for highly nonlinear, nonconvex, nonsmooth, multi-agent systems. We prove contraction and robustness to latency for multi-agent, nonconvex problems, showing applicability to real-world cons…

arXiv:2608.27584v1 Announce Type: new Abstract: Autonomous systems rely on extracting information from light, yet remain brittle in extreme environments, from nighttime navigation to high-speed robotics. Conventional sensors aggregate photons over fixed exposures, imposing trade-offs between sensitivity, dynamic range, and temporal resolution that degrade perception when photons are scarce or dynamics are rapid. Quanta sensors detect individual photons, but their streams exceed real-time compute and latency budgets by orders of magnitude. Here we introduce $\textit{probabilistic events}$, a computational primitive for real-time quanta perception from individual photon detections. By computing the posterior over the time since the last intensity change, we represent photon streams as recur…

arXiv:2608.27466v1 Announce Type: new Abstract: Web content extraction is essential for reliable LLM data pipelines, yet existing methods often struggle to jointly satisfy accuracy, scalability, and adaptability. General-purpose extractors can be applied broadly, but they are often brittle on publisher-specific layouts and richer extraction targets such as metadata, images, and tables. Direct LLM-based extraction offers greater flexibility, but incurs substantial cost and latency at scale, while manually engineered publisher-specific parsers can achieve high accuracy but require substantial human effort to build and maintain. We introduce PACE, an agentic framework for learning publisher-specific extraction configurations from representative pages and user requirements. During training, P…

arXiv:2608.27460v1 Announce Type: new Abstract: Large output embedding matrices create a significant memory bandwidth bottleneck during autoregressive decoding, especially for compact LLMs with large multilingual vocabularies. We reformulate the output projection followed by top-k token selection as a maximum inner product search over token embeddings and replace the dense vocabulary projection with an HNSW-based vector index. The resulting output head retrieves only a small candidate set of high-scoring tokens and can be integrated into existing decoding pipelines by scattering retrieved logits into a sparse full-vocabulary tensor. On CPU inference with Gemma 3, Llama 3.2, and Qwen 3 models, our method substantially accelerates the output projection and improves end-to-end batch-size-one…

arXiv:2608.27513v1 Announce Type: new Abstract: Softmax attention stores key and value vectors for every preceding token, causing inference memory to grow with sequence length. Recent language models incorporating Gated DeltaNet (GDN) or Kimi Delta Attention (KDA) reduce this cost by replacing the KV cache in most layers with fixed-size recurrent states. However, these recurrent states are commonly stored in FP32 and consume substantial GPU memory; their updates are memory-bandwidth bound and contribute significantly to decoding latency. To our knowledge, we are the first to study post-training quantization of recurrent states in GDN and KDA based language models. We find that uniform quantization provides a poor accuracy--storage trade-off: INT8 and FP8 already degrade accuracy on comple…

arXiv:2608.27512v1 Announce Type: new Abstract: Post-training quantization is often treated as a semantically neutral optimization for edge deployment of Large Language Models. When a full-precision source checkpoint is evaluated and quantization is applied downstream without equivalent re-evaluation, this workflow creates a structural validation--deployment gap: because quantization is a many-to-one mapping over parameter space, source-precision certification does not guarantee behavioral equivalence in the deployed configuration. We formalize this gap through Quantization Behavioral Equivalence Classes (QBECs) and prove that QBEC membership does not imply behavioral equivalence, providing a theoretical basis for quantization-triggered backdoor attacks. Building on a three-stage adversar…

Voice agents fail on latency long before they fail on intelligence. Time to first token is the metric most teams use to choose an inference API, and it is the right starting point and the wrong stopping point. This benchmark works through every layer of the voice stack — LLM, speech-to-text, text-to-speech, and speech-to-speech — using figures verified against primary sources on August 30, 2026, with each number labeled as independently measured, vendor-published, or vendor-measured on its own product. The post Lowest-Latency Inference APIs for Voice and Realtime Agents: A Time to First Token TTFT-First Benchmark appeared first on MarkTechPost.

This article walks through what each technique actually does, why skipping them costs real money and real latency, and then gets hands-on with five specific methods people are running in production right now.

arXiv:2608.26496v1 Announce Type: new Abstract: Navigation in unknown environments to find unforeseen objects has become increasingly feasible with capable vision and language foundation models. However, these models also introduce non-negligible inference latency, which becomes an important concern when agents must operate continuously in the real world. Most state-of-the-art methods are still developed in synchronous simulators, where the environment waits for the agent to act and inference time is effectively free. As a result, agents are often designed around the sequential execution of perception, reasoning, and action, with little regard for time constraints. Under real-time execution, where wall-clock time counts towards the task budget, the inefficiencies of these architectures be…

arXiv:2608.26383v1 Announce Type: new Abstract: Autonomous robots performing laboratory tasks depend on 3D reconstruction pipelines that can turn raw camera streams into actionable object representations within the latency budget of a physical control loop. Neural 3D reconstruction methods have demonstrated high-quality view synthesis, but their real-time viability across the compute platforms on which laboratory robots actually run remains poorly characterized. In this work, we present a systematic compute-platform benchmark of neural 3D reconstruction methods, evaluating NeRF and 3D Gaussian Splatting training and rendering on GPU-enabled computing devices ranging from single-board computers to server-class nodes, and place Meta's SAM3D single-image reconstruction on the same axes to qu…

arXiv:2608.26112v1 Announce Type: new Abstract: Speculative decoding accelerates large language model inference through a draft-then-verify paradigm. Building on this, tree-structured methods improve inference by organizing proposals into multiple candidate paths, increasing the accepted length. However, existing tree-structured methods use a single drafter for all drafting steps, creating a dilemma: a smaller drafter is fast but yields lower-quality trees, whereas a larger drafter improves tree quality but suffers from high latency. To address this, we propose TreeGraft, a multi-drafter framework in which drafters of different costs jointly construct a shared draft tree. TreeGraft uses the stronger drafter to rescore candidates by updating scores assigned by the weaker drafter, reselect…

Serving automatic speech recognition (ASR) models at scale is costly when each request uses only a fraction of a GPU. Learn how NVIDIA CUDA Multi-Process Service (MPS) with NVIDIA Triton Inference Server on Amazon EC2 GPU instances cuts GPU infrastructure by 75% while holding sub-second latency at 92.1 requests per second per GPU.

arXiv:2608.25140v1 Announce Type: new Abstract: Existing approaches to building line-level Arabic handwritten-text-recognition (HTR) training data either rely on fully manual annotation, which does not scale, or on automatic OCR-to-reference alignment methods not yet extended to multi-script, two-zone (main-plus-margin) manuscript layouts with a provable correctness guarantee. We present RefLAM (Reference-grounded Line Annotation for Manuscripts), a pipeline converting manuscript page images and clean transcriptions into validated, line-level ground truth without sacrificing human oversight. RefLAM couples a deep-learning page-segmentation model with a multimodal large language model (MLLM) for structured OCR and a diacritic-agnostic fuzzy alignment engine that grounds each OCR line in a…

arXiv:2608.25068v1 Announce Type: new Abstract: Depth pruning removes entire Transformer blocks to reduce the inference cost of large language models, but disrupts the hidden-state distributions expected by downstream layers, leading to significant accuracy loss. We introduce SHIFT-LLM, a training-free post-pruning correction framework that inserts a Linear Residual Adapter (LRA) at each pruning site. Each LRA preserves the identity pathway of the original residual block and adds a lightweight affine residual correction. This correction is calibrated via closed-form least-squares regression on a small held-out set, without gradient computation, to approximate the missing residual update produced by the pruned block. Together with the preserved identity pathway, the resulting LRA output ap…

arXiv:2608.24935v1 Announce Type: new Abstract: Accurate identification of early-stage apple fruitlet anatomical structures, including the calyx, fruitlet body, and peduncle, is essential for robotic thinning, crop-load management, and other precision orchard operations. This study presents a lightweight multimodal vision-language framework that adapts TinyCLIP for fine-grained fruitlet anatomy classification in complex orchard environments. A dataset of 600 high-resolution RGB images collected from Scilate and Scifresh apple orchards was converted into 224 x 224 image patches and annotated for three anatomical classes. Domain-specific language prompts, such as ``a photo of a class,'' were used to guide multimodal alignment between orchard imagery and horticultural structures. A sliding-w…

arXiv:2608.24945v1 Announce Type: new Abstract: Recent years have witnessed remarkable achievements of Large Language Models (LLMs) in multiple domains, while the excessive resource requirements of LLMs hinder the deployment on resource-constrained devices. Although model quantization stands out as an effective approach, conventional quantization approaches typically incur severe performance degradation due to uniform bit-width or simple heuristic sensitivity evaluation. In this paper, we propose a novel Fisher information-based Adaptive Mixed Precision Weight Quantization approach, i.e., FAMPWQ, which performs layer-adaptive weight quantization for effective LLM inference on commodity GPUs. First, we propose a system model with a novel Fisher information metric to measure the layer-wise…

arXiv:2608.24938v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) models scale capacity for strong quality while keeping per-token compute bounded through sparse expert activation. Yet low-latency MoE serving is increasingly challenging, because it spans two inference phases with fundamentally different bottlenecks: prefill is dominated by token-wise expert computation, whereas decode is constrained by memory traffic from the batch-wise activated expert set. However, existing training-free acceleration methods optimize only a single resource proxy, either the experts each token executes or the experts a batch activates, and either discard the excluded experts' contribution or leave it only implicitly approximated. In this paper, we propose ExFold, a unified training-free expert-fol…

arXiv:2608.24936v1 Announce Type: new Abstract: We present GreenLeaf Law Embed Tiny, a 0.6B parameter embedding model for legal domain retrieval. GreenLeaf-Tiny achieves 75.11% on the Massive Legal Embedding Benchmark (MLEB) and 64.38% on MTEB(Law, v1),demonstrating competitive performance among models under 1B parameters. Our approach combines a two-stage training pipeline that first distills knowledge from a larger teacher model into a compact student architecture, then applies domain-specific fine-tuning with hard negative mining; a carefully curated dataset of 3.4 million query-passage pairs, including 150,000 human-curated samples across diverse legal jurisdictions; and an efficient inference architecture supporting multiple quantization levels (BF16, INT8, binary) enabling deploymen…

Z.ai has released GLM-5.3-Flash, the first natively multimodal model in the GLM-5 series — a 320B-total / 18B-active MoE with a 1,048,576-token context window, MIT-licensed weights on Hugging Face, and API pricing at $0.15/M input and $0.50/M output. It scores 84.3 on Terminal-Bench 2.1 and 63.4 on DeepSWE v1.1, using hybrid KDA linear plus NoPE sparse MLA attention to cut attention compute ~3× and KV cache 4.4× versus GLM-5.3. The post Z.ai Releases GLM-5.3-Flash: A 320B-A18B Natively Multimodal MoE With a 1M-Token Context appeared first on MarkTechPost.

Learn proven strategies to speed up your AI agent: reduce latency, optimize LLM calls, enable parallelism, and improve UX. Expert tips from LangChain.

arXiv:2608.23831v1 Announce Type: new Abstract: While reinforcement learning (RL) allows generalist robot policies to continually improve during deployment, the large model size of modern generalist policies, such as VLAs, poses a fundamental obstacle to effective RL improvement. In particular, their severe inference latency---which can lead to pauses or jerky movements---can alter the effective environment dynamics and, if not correctly accounted for, break the Markov assumption that RL relies on, causing standard RL algorithms to fail completely. In this work, we introduce a latency-aware framework, Asynchronous RL with Intermediate Information (ARLI), that enables RL-based improvement of generalist policies under inference delays. Our framework builds on asynchronous inference approach…

arXiv:2608.23634v1 Announce Type: new Abstract: Many few-shot adaptation methods for vision-language models classify with a convex combination of the zero-shot text prototype and the mean of the K labelled image features, with a single blending ratio routinely tuned on held-out labels, often on the test set itself. We ask what the family's own bias-variance justification invites: what is the right ratio, can it be estimated without validation data, and is finding it where the performance is? First, the ratio minimising prototype mean-squared error has a closed form whose support-set plug-in is exactly a positive-part James-Stein coefficient shrinking towards the text prototype. Across 4,800 cells (ten datasets, five backbones including SigLIP, five shot counts, five seeds, four prompt tie…

arXiv:2608.23794v1 Announce Type: new Abstract: Mixture-of-Experts (MoE) scales language models by routing each input through a small set of independently parameterized experts. We show that copying this design into convolutional networks fails for a structural reason: parallel convolutional experts that read the same input channels learn nearly identical filters. We therefore move the expert axis from operator duplication to channel selection. We introduce Mixture of Channel Experts (MoCE), a structured sparse channel-mixing layer, inspired by MoE, that replaces pointwise (1x1) channel-reduction projections. In MoCE, an expert is a single output channel with a learned sparse support of k << C input channels. The selected channels are combined by a softmax whose temperature is predicted p…

Model cards report quality under server-class, full-precision conditions. Those numbers rarely predict how the same model behaves on a phone. This week, Liquid AI released Pipette. It is an open-source platform for benchmarking foundation models on edge devices, built in partnership with Artificial Analysis as an independent methodology validator. Pipette treats on-device behavior as a […] The post Liquid AI Open-Sources Pipette: A Reproducible Benchmarking Suite That Measures On-Device Models, Quantization, Runtime and Hardware Together appeared first on MarkTechPost.

OpenAI says its new AI chip, Jalapeño, completes tasks more efficiently and returns responses faster than other AI systems, according to a blog post published on Tuesday. During a briefing with reporters, OpenAI hardware vice president Richard Ho said Jalapeño offers the "best of both worlds" with lower latency and higher throughput, as AI systems typically "have to make a trade-off between the two." First introduced in June, Jalapeño is an Application-Specific Integrated Circuit (ASIC) made in partnership with Broadcom. It's designed for AI inference - the process of running a trained AI model to complete a task or deploy an agent. To mea … Read the full story at The Verge.

This article is brought to you by Ollobot. From about 2017, individuals began to truly connect with the initial wave of companion robots. These devices had personality, moved around, joked, and answered when you spoke to them. Most early companion robots, however, were still limited by simple voice-command interactions and narrow functionality. Once the novelty wore off, many ended up sitting unused on shelves. As some of those companies went out of business and turned off their servers, many owners likened it to losing a pet. What Ollobot describes as “gentle intelligence” is a useful way to think about where the serious work in this category is going. Not toward more powerful assistants, but toward more present ones. The problem companion robots were trying to solve Loneliness is not a…

Jalapeño is a custom inference chip from OpenAI that delivers faster, more power-efficient AI inference, with higher throughput and lower latency for modern models.

arXiv:2608.21393v1 Announce Type: new Abstract: Running large language models inside enterprise environments has always bumped up against a practical wall: the data lives in one place, the AI horsepower sits somewhere else, and moving sensitive records between the two creates real headaches around latency, security, and regulatory exposure. IBM's Spyre accelerator PCIe inference card built for LinuxONE and the broader IBM Z family changes that equation. In this paper we lay out a six-subsystem RAG architecture that runs entirely on IBM LinuxONE, using Spyre for generative inference, the Telum II on-chip accelerator for lightweight classification tasks, and Red Hat OpenShift for container orchestration. Every piece of the pipeline from query intake through vector retrieval, prompt assembly…

arXiv:2608.21375v1 Announce Type: new Abstract: Heterogeneous agentic retrieval-augmented generation (RAG) systems increasingly orchestrate external APIs, internal databases, vector stores, and graph stores. Exposing all tool descriptions to an LLM agent, or selecting tools only by vector similarity, causes two costly failures: over-fetching, which increases payload size, token use, and latency, and under-fetching, which omits fields needed to answer the query. We present SchemaRouter, a lightweight routing layer that represents tools, endpoints, parameters, response fields, domain concepts, units, provenance, and license policies as a schema graph. Given a query, SchemaRouter emits an executable tool plan specifying which tools to call and which fields to retrieve. A small LLM extracts i…