model inference
30 articles about model inference in AI news
Open-Weight 1T Model Inference Margins Hit 88% on Rented GPUs
Renting a 128 GPU cluster to serve a 1T open model yields ~88% margin on tokens sold at $0.002/1K, exposing a structural arbitrage over proprietary APIs.
Groq's LPU Inference Engine Demonstrates 500+ Token/s Performance on Llama 3.1 70B
Groq's Language Processing Unit (LPU) inference engine achieves over 500 tokens/second on Meta's Llama 3.1 70B model, demonstrating significant performance gains for large language model inference.
Colibri Runs 744B-Parameter Model on 25GB RAM, No GPU
Colibri claims to run a 744B-parameter model on 25GB RAM without GPU, but lacks evidence. If true, it could democratize large-model inference.
Apple Core AI Runs Models On-Device, Zero Server Calls
Apple launched Core AI for on-device model inference on Apple silicon. Zero server calls, supports Qwen, Mistral, SAM3 across devices.
OpenAI Cuts Inference Costs by Half on Some Models
OpenAI cut inference costs by 50%+ on some models for logged-out ChatGPT users, per The Information. The move reduces operational expenses.
Perplexity Claims 3x Blackwell Inference Throughput for 70B Models
Perplexity AI claims 3x inference throughput for 70B models on Nvidia Blackwell GPUs via FP4 and custom scheduling. The gain exceeds Nvidia's own 2x marketing claim.
Switchcraft Router Cuts Agentic AI Inference Cost 84%, Matches Top Model
Switchcraft, a DistilBERT-based model router for agentic tool calling, achieves 82.9% accuracy while cutting inference cost by 84%, saving over $3,600 per million queries.
Dflash with Continuous Batch Inference Teased for Draft Models
A developer teased the upcoming release of 'Dflash' with continuous batch inference, targeting current text-only draft models used in speculative execution to speed up LLM inference.
mlx-vlm v0.4.2 Adds SAM3, DOTS-MOCR Models and Critical Fixes for Vision-Language Inference on Apple Silicon
mlx-vlm v0.4.2 released with support for Meta's SAM3 segmentation model and DOTS-MOCR document OCR, plus fixes for Qwen3.5, LFM2-VL, and Magistral models. Enables efficient vision-language inference on Apple Silicon via MLX framework.
The Hidden Cost of Mixture-of-Experts: New Research Reveals Why MoE Models Struggle at Inference
A groundbreaking paper introduces the 'qs inequality,' revealing how Mixture-of-Experts architectures suffer a 'double penalty' during inference that can make them 4.5x slower than dense models. The research shows training efficiency doesn't translate to inference performance, especially with long contexts.
mlx-vlm v0.5.0 Adds Continuous Batching, Distributed Inference for Apple Silicon
mlx-vlm v0.5.0 adds continuous batching, speculative decoding, and distributed inference for Apple Silicon. The release supports Qwen3.5, Kimi K2.5, Gemma 4 video, and new models with 21 contributors.
DeepSeek-V4 Ported to MLX for Apple Silicon Inference
A developer has ported DeepSeek-V4 to Apple's MLX framework, allowing the large language model to run on Apple Silicon Macs. Early results show functional inference with room for optimization.
SemiAnalysis: NVIDIA's Customer Data Drives Disaggregated Inference, LPU Surpasses GPU
SemiAnalysis states NVIDIA's direct customer feedback is leading the industry toward disaggregated inference architectures. In this model, specialized LPUs can outperform GPUs for specific pipeline tasks.
Prefill-as-a-Service Paper Claims to Decouple LLM Inference Bottleneck
A research paper proposes a 'Prefill-as-a-Service' architecture to separate the heavy prefill computation from the lighter decoding phase in LLM inference. This could enable new deployment models where resource-constrained devices handle only the decoding step.
Cursor AI Claims 1.84x Faster MoE Inference on NVIDIA Blackwell GPUs
Cursor AI announced a rebuilt inference engine for Mixture-of-Experts models on NVIDIA's new Blackwell GPUs, resulting in a claimed 1.84x speedup and improved output accuracy.
Image Prompt Packaging Cuts Multimodal Inference Costs Up to 91%
A new method called Image Prompt Packaging (IPPg) embeds structured text directly into images, reducing token-based inference costs by 35.8–91% across GPT-4.1, GPT-4o, and Claude 3.5 Sonnet. Performance outcomes are highly model-dependent, with GPT-4.1 showing simultaneous accuracy and cost gains on some tasks.
Apple M5 Max NPU Benchmarks 2x Faster Than Intel Panther Lake NPU in Parakeet v3 AI Inference Test
A leaked benchmark using the Parakeet v3 AI speech recognition model shows Apple's next-generation M5 Max Neural Processing Unit (NPU) delivering double the inference speed of Intel's competing Panther Lake NPU. This real-world test provides early performance data in the intensifying on-device AI hardware race.
Nvidia Claims MLPerf Inference v6.0 Records with 288-GPU Blackwell Ultra Systems, Highlights 2.7x Software Gains
MLCommons released MLPerf Inference v6.0 results, introducing multimodal and video model tests. Nvidia set records using 288-GPU Blackwell Ultra systems and achieved a 2.7x performance jump on DeepSeek-R1 via software optimizations alone.
Fractal Emphasizes LLM Inference Efficiency as Generative AI Moves to Production
AI consultancy Fractal highlights the critical shift from generative AI experimentation to production deployment, where inference efficiency—cost, latency, and scalability—becomes the primary business constraint. This marks a maturation phase where operational metrics trump model novelty.
TTQ: A New Framework for On-the-Fly Quantization of LLMs at Inference Time
Researchers propose TTQ, a test-time quantization method that compresses large language models dynamically during inference. It uses efficient online calibration to adapt to any prompt, aiming to solve domain-shift issues and accelerate inference without retraining.
Why Companies End Up Using Triton Inference Server: A Simple Case Study
A case study explains the common journey from a simple ML experiment to a production system requiring a robust inference server like NVIDIA's Triton, highlighting its role in managing multi-model, multi-framework deployments at scale.
Kimi's Selective Layer Communication Improves Training Efficiency by ~25% with Minimal Inference Overhead
Kimi has developed a method that replaces uniform residual connections with selective information routing between layers in deep AI models. This improves training stability and achieves ~25% better compute efficiency with negligible inference slowdown.
IonRouter Emerges as Cost-Efficient Challenger to OpenAI's Inference Dominance
YC-backed Cumulus Labs launches IonRouter, a high-throughput inference API that promises to slash AI deployment costs by optimizing for Nvidia's Grace Hopper architecture. The service offers OpenAI-compatible endpoints while enabling teams to run open-source or fine-tuned models without cold starts.
AWS Expands Claude AI Access Across Southeast Asia with Global Cross-Region Inference
Amazon Bedrock now offers Global Cross-Region Inference for Anthropic's Claude models in Thailand, Malaysia, Singapore, Indonesia, and Taiwan. This enables enterprise customers to access Claude Opus 4.6, Sonnet 4.6, and Haiku 4.5 through a resilient, distributed architecture designed for high-throughput AI applications.
China Blank-Checks Huawei for Inference Chip Push
China is giving Huawei a blank check for inference chips, per a tweet by Natolambert on February 3, 2026, shifting focus from training to inference.
LLMs Learn to Switch Reasoning Effort at Inference Time
@rasbt explains how LLMs switch reasoning effort using inference-time methods and training, potentially cutting token usage by 30–50% on simple queries.
Hugging Face weekly papers: Monotonic inference policy overtakes training optimization
Hugging Face's top papers July 6-12 include a paper arguing monotonic inference policies are the true LLM RL objective, and Vidu S1 for real-time interactive video generation.
DeepSeek, Zhipu AI Build Custom Inference Chips to Cut GPU Dependency
DeepSeek and Zhipu AI are developing custom inference chips to cut GPU costs. China's domestic chip budget share hit 46% in July 2026.
ZML releases free LLM inference server supporting Nvidia
ZML released LLMD, a free inference server for LLMs supporting Nvidia, AMD, Google TPU, Apple Metal, and Intel Arc, aiming to reduce AI costs and break vendor lock-in.
Alibaba's MIPI fixes LLM training-inference mismatch with direct RL
Alibaba's MIPI directly optimizes inference policy, fixing the mismatch in LLM post-training via the MIPU framework.