Skip to content
gentic.news — AI News Intelligence Platform
Connecting to the Living Graph…

speculative decoding

30 articles about speculative decoding in AI news

DeepSeek DSpark: Speculative Decoding Unifies Parallel Gen, Adaptive Verification

DeepSeek released DSpark, a speculative decoding framework unifying parallel generation with adaptive verification. No benchmarks disclosed yet; the approach targets inference latency and throughput.

90% relevant

JetSpec hits 1,000 t/s on Qwen-8B with speculative decoding

JetSpec achieves 1,000 t/s on Qwen-8B with a B200 GPU, claiming superiority over prior speculative decoding methods, but lacks independent verification.

89% relevant

NVIDIA NeMo RL Speculative Decoding: 1.8× Rollout Speed at 8B

NVIDIA's NeMo RL speculative decoding achieves 1.8× rollout speedup at 8B and projects 2.5× at 235B, cutting RL training time by over half.

72% relevant

PayPal Cuts LLM Inference Cost 50% with EAGLE3 Speculative Decoding on H100

PayPal engineers applied EAGLE3 speculative decoding to their fine-tuned 8B-parameter commerce agent, achieving up to 49% higher throughput and 33% lower latency. This allowed a single H100 GPU to match the performance of two H100s running NVIDIA NIM, cutting inference hardware cost by 50%.

90% relevant

DFlash Brings Speculative Decoding to Apple Silicon via MLX

DFlash, a new open-source project, implements speculative decoding for large language models on Apple Silicon using the MLX framework, reportedly delivering up to 2.5x speedup on an M5 Max.

85% relevant

NVIDIA's Kimi-K2.5 Eagle Head: Supercharging Moonshot's Reasoning with Speculative Decoding

NVIDIA has released the Kimi-K2.5 Eagle head on Hugging Face, implementing Eagle-3 speculative decoding to dramatically accelerate inference for Moonshot's reasoning models. This breakthrough promises blazing-fast performance while maintaining accuracy.

89% relevant

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.

87% relevant

Paper Details Full-Stack MFM Acceleration: Quant, Spec Decode, HW Co-Design

A research paper details a full-stack approach for accelerating multimodal foundation models, combining hierarchy-aware mixed-precision quantization, structural pruning, speculative decoding, model cascading, and a specialized hardware accelerator. Demonstrated on medical and code generation tasks.

72% relevant

Claude Code's Secret Efficiency Hack

Claude Code leverages speculative decoding to reduce LLM energy use by 100x. Learn how this built-in optimization makes your coding faster and cheaper.

85% relevant

Nebius AI's LK Losses: A Breakthrough in Making Large Language Models Faster and More Efficient

Nebius AI has introduced LK Losses, a novel training objective that directly optimizes acceptance rates in speculative decoding. This approach achieves 8-10% efficiency gains over traditional methods, potentially revolutionizing how large language models are deployed.

85% relevant

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.

85% relevant

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.

85% relevant

SWE-Pruner Pro Saves 39% Tokens by Reading LLM Hidden States

SWE-Pruner Pro saves up to 39% tokens on coder LLMs by reading keep-or-prune signals from hidden states, maintaining task quality without external heuristics.

87% relevant

7 AI Agent Cost Optimization Strategies That Cut LLM Bills by Up to 90%

The source outlines seven cost optimization strategies for AI agents, including prompt compression and model routing, that can reduce LLM bills by up to 90%. This matters for retail and luxury brands deploying AI at scale where inference costs can become prohibitive.

69% relevant

Hugging Face Papers: 35B Agent Matches Trillion-Parameter Performance

Hugging Face Daily Papers featured eight AI papers, including Orca (world model), Dockerless (62% SWE-bench), and a 35B agent matching trillion-parameter performance.

85% relevant

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.

91% relevant

NVIDIA Blackwell Cuts DeepSeek V4 Token Costs 5x in One Month

NVIDIA claims Blackwell inference stack cut DeepSeek V4 token costs 5x in one month, per a newly published report shared by @rohanpaul_ai.

100% relevant

Grouped Query Experts cuts long-context attention cost 44%

GQE speeds long-context attention prefill 1.7–1.8× by routing tokens to 9 of 16 query heads, matching baseline accuracy at 56.04.

85% relevant

DeepSeek Raises $7B, Ends No-Funding Pledge, Doubles Staff

DeepSeek raised $7B, abandoning its no-funding pledge, to double headcount and launch a coding agent team competing with Claude Code.

100% relevant

llada.cpp Cuts LLaDA-8B Latency 17-42x on Mobile NPU

llada.cpp, the first NPU-aware dLLM inference framework, cuts LLaDA-8B latency 17-42x on smartphones, enabling real-time on-device generation.

84% relevant

OpenRouter Fusion API Claims Fable-Level IQ at Half the Cost

OpenRouter's Fusion API routes queries across providers to match Fable-level intelligence at half the cost, per company claims. No third-party benchmarks disclosed.

87% relevant

MiniMax M3: Sparse Attention, 1M Context, Multimodal via Together

MiniMax M3 uses sparse attention for 1M context and multimodality, with Together AI serving fast inference.

95% relevant

Median Coding Agent Hits 96k Input Tokens, Rewriting Inference Economics

SemiAnalysis found median coding agent uses 96k input tokens from 432k requests, shifting inference cost focus from output to context.

95% relevant

Compute Shortage to Split AI Market: Rich Get Agents, Poor Get Chatbots

Mollick warns compute shortage makes agents expensive while chatbots cheapen, splitting AI market by company resources.

75% relevant

Qwen 3.6 27B Hits 34 tok/s on M5 Max MacBook Pro

Qwen 3.6 27B hits 34 tok/s on M5 Max MacBook Pro with 90% acceptance rate, per @rohanpaul_ai. Shows viable local LLM inference on Apple Silicon.

75% relevant

LASAR Cuts Latent Reasoning Steps in Half for GenRec at 20x Speedup Over CoT

LASAR nearly halves latent reasoning steps and achieves 20x speedup over explicit CoT in generative recommendation, outperforming baselines on three datasets.

80% relevant

Google Gemma 4: 3x Faster Inference with MTP Drafters

Google's Gemma 4 claims up to 3x faster inference via MTP drafters, but released no benchmark numbers or architectural details.

85% relevant

RedParrot: Semantic Caching Speeds Up NL-to-DSL for Business Analytics by

Xiaohongshu researchers propose RedParrot, a framework that caches normalized structural patterns of natural language queries to bypass expensive LLM pipelines, achieving 3.6x speedup and 8.26% accuracy improvement on enterprise datasets.

84% relevant

Sam Altman: AI inference costs dropped 1000x from o1 to GPT-5.4

Sam Altman stated AI inference costs for solving a fixed hard problem dropped ~1000x from o1 to GPT-5.4 in ~16 months, crediting cross-layer engineering optimizations, not a single breakthrough.

85% relevant

TACO Framework Cuts Agent Token Overhead 10% via Self-Evolving Compression

Researchers introduced TACO, a framework that enables terminal agents to automatically discover and refine context compression rules from their own interaction trajectories. This approach cuts token overhead by approximately 10% on benchmarks like TerminalBench and SWE-Bench Lite while preserving task accuracy.

87% relevant