无损压缩AI智能体消息，token直降36%，成本优化利器，开源可即用。
Article URL: https://github.com/reh8n/a2acompress Comments URL: https://news.ycombinator.com/item?id=49281816 Points: 4 # Comments: 0
让全模态大模型告别冗余Token，解耦压缩新思路，先分配后排序，推理效率可期！
arXiv:2608.01665v1 Announce Type: new Abstract: Token compression in OmniLLMs is typically posed as a single saliency-ranking problem: score each mult…
针对全模态大模型，提出基于技能驱动的token压缩预算分配方法OmniDelta，提升效率保留关键信息。
arXiv:2607.25669v1 Announce Type: new Abstract: Emerging Omni-modal Large Language Models (OmniLLMs) enable unified understanding of text, audio, and …
多模态大模型有惊喜：砍掉一半视觉token反而更鲁棒？最新研究揭秘token压缩的抗干扰机制。
arXiv:2607.22716v1 Announce Type: cross Abstract: In this paper, we show for the first time that visual token pruning enhances the robustness of Multi…
多模态大模型也能「压缩记忆」？这篇论文提出在不牺牲理解能力的前提下，高效压缩视觉与文本token，让全模态LLM更轻更快。
arXiv:2607.21179v1 Announce Type: new Abstract: The goal of this paper is to reduce the input token cost of Omni-modal large language models (Omni-LLM…
一小时级音视频理解新突破，检索启发式token压缩让全模态大模型兼顾效率与精度。
arXiv:2606.24286v1 Announce Type: new Abstract: Multimodal Large Language Models have achieved remarkable progress in short-form audio-video understan…
用1%的token复现全上下文回忆，开源基准实测有效，直击长上下文“lost in the middle”痛点。
Article URL: https://github.com/compresh/compresh-benchmarks/blob/main/epbench/WRITEUP.md Comments URL: https://news.ycombinator.com/item?id=48641509 …
给AI Agent的上下文做瘦身：JSON数据最高省95%token，代码场景省15-20%，还支持代理/MCP接入，性能可观。
Article URL: https://github.com/headroomlabs-ai/headroom Comments URL: https://news.ycombinator.com/item?id=48627072 Points: 1 # Comments: 0
工具输出压缩能否真正省token？效率与轮次的博弈催生LLM工具新生态。
Many projects and addons exists to reduce the verbosity on standard bash / git /npm etc. Commands that agents pass regularly as tools to LLMs. (eg. rt…
AI自我进化新突破：将搜索代码从1260行压减至75行，准确率提升7.7%，token消耗减少4倍。
Article URL: https://github.com/001TMF/harness-forge Comments URL: https://news.ycombinator.com/item?id=48526233 Points: 1 # Comments: 0
一个能一键安装 token 压缩引擎的开源工具，让 AI 工具更轻量高效。
the is nothing to explain about it , if I get some ups on my message I will make the tool free for all find it at https://slipstream.li Or skip the ma…
比压缩算法好2倍：Lore用智能记忆管理让AI编码代理告别68分钟/天的重复解释，总召回率提升至2.6倍。
Article URL: https://withlore.ai/ Comments URL: https://news.ycombinator.com/item?id=48464573 Points: 4 # Comments: 0
实时压缩代码上下文，降低LLM API成本50-80%，可直接嵌入的代理工具。
Article URL: https://github.com/borhen68/TokenTamer Comments URL: https://news.ycombinator.com/item?id=48458633 Points: 1 # Comments: 1
AI代理读取内容前自动压缩，最高减少90% token消耗，实现成本断崖式下降。
Article URL: https://pypi.org/project/headroom-ai/ Comments URL: https://news.ycombinator.com/item?id=48349275 Points: 1 # Comments: 0
CVPR 2026最新研究：通过早期Token压缩实现快速视频理解，颠覆传统视频处理效率瓶颈。
arXiv:2605.30010v1 Announce Type: new Abstract: Video large language models (Video-LLMs) have demonstrated strong capabilities in video understanding …
通过压缩词空间加速约束解码，让LLM输出精准服从语法规则，显著提升效率。
arXiv:2605.29986v1 Announce Type: new Abstract: To guarantee that an LLM's outputs conform to a specified structure, context-free grammar (CFG) decodi…
GitHub开源项目，让LLM应用拥有长期记忆，同时将输入token平均削减68%，大幅降低API成本。
Article URL: https://github.com/Tem-Degu/streetai-memory Comments URL: https://news.ycombinator.com/item?id=48249509 Points: 1 # Comments: 0
提出高效视觉编码器，解决Video LLM长视频中视觉token爆炸难题，突破帧扩展瓶颈。
arXiv:2605.17260v1 Announce Type: new Abstract: The fundamental challenge in scaling Video Large Language Models (Video LLMs) to long-form video lies …
提出频域残差压缩方法，大幅减少视频MLLM的token数量，高效且不损失性能。
arXiv:2605.16366v1 Announce Type: new Abstract: Video MLLMs face a persistent tension between spatial fidelity and temporal coverage: preserving fine-…