[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"project-77605":3},{"id":4,"name":5,"fullName":6,"owner":7,"repo":5,"description":8,"homepage":9,"htmlUrl":10,"language":11,"languages":10,"totalLinesOfCode":10,"stars":12,"forks":13,"watchers":14,"openIssues":15,"contributorsCount":16,"subscribersCount":16,"size":16,"stars1d":17,"stars7d":18,"stars30d":19,"stars90d":16,"forks30d":16,"starsTrendScore":20,"compositeScore":21,"rankGlobal":10,"rankLanguage":10,"license":22,"archived":23,"fork":23,"defaultBranch":24,"hasWiki":25,"hasPages":23,"topics":26,"createdAt":10,"pushedAt":10,"updatedAt":44,"readmeContent":45,"aiSummary":46,"trendingCount":16,"starSnapshotCount":16,"syncStatus":47,"lastSyncTime":48,"discoverSource":49},77605,"parallax","GradientHQ\u002Fparallax","GradientHQ","Parallax is a distributed model serving framework that lets you build your own AI cluster anywhere","",null,"Python",1310,139,25,20,0,1,9,19,8,19.44,"Apache License 2.0",false,"main",true,[27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43],"blackwell","chatbot","decentralized-inference","deepseek","distributed-systems","glm","kimi","large-language-models","llama","llm","llm-serving","minimax","oss-gpt","python","pytorch","qwen","transformer","2026-06-12 02:03:43","\u003Cdiv align=\"center\">\n  \u003Cp align=\"center\">\n    \u003Cimg src=\"docs\u002Fimages\u002Fparallax.png\" width=\"720\">\n    \u003Cdiv align=\"center\">\n      \u003Cp style=\"font-size: 1.3em; font-weight: 600; margin-bottom: 10px;\">Trusted by Partners\u003C\u002Fp>\n      \u003Cimg src=\"docs\u002Fimages\u002Fsglang.png\" alt=\"SGLang\" height=\"28\" style=\"margin: 0 20px;\">\n      \u003Cimg src=\"docs\u002Fimages\u002Fvllm.png\" alt=\"vLLM\" height=\"30\" style=\"margin: 0 20px;\">\n      \u003Cimg src=\"docs\u002Fimages\u002Fqwen.avif\" alt=\"Qwen\" height=\"30\" style=\"margin: 0 20px;\">\n      \u003Cimg src=\"docs\u002Fimages\u002Fdeepseek.png\" alt=\"DeepSeek\" height=\"30\" style=\"margin: 0 20px;\">\n      \u003Cimg src=\"docs\u002Fimages\u002Fkimi.png\" alt=\"Kimi\" height=\"30\" style=\"margin: 0 20px;\">\n      \u003Cimg src=\"docs\u002Fimages\u002Fminimax.png\" alt=\"Minimax\" height=\"30\" style=\"margin: 0 10px;\">\n      \u003Cimg src=\"docs\u002Fimages\u002Fzai.svg\" alt=\"ZAI\" height=\"30\" style=\"margin: 0 10px;\"\u002F>\n    \u003C\u002Fdiv>\n  \u003C\u002Fp>\n\n[![license](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Flicense\u002FGradientHQ\u002Fparallax.svg)](https:\u002F\u002Fgithub.com\u002FGradientHQ\u002Fparallax\u002Ftree\u002Fmain\u002FLICENSE)\n[![issue resolution](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fissues-closed-raw\u002FGradientHQ\u002Fparallax)](https:\u002F\u002Fgithub.com\u002FGradientHQ\u002Fparallax\u002Fissues)\n[![open issues](https:\u002F\u002Fimg.shields.io\u002Fgithub\u002Fissues-raw\u002FGradientHQ\u002Fparallax)](https:\u002F\u002Fgithub.com\u002FGradientHQ\u002Fparallax\u002Fissues)\n\n\u003Ca href=\"https:\u002F\u002Fwww.producthunt.com\u002Fproducts\u002Fparallax-by-gradient?embed=true&utm_source=badge-top-post-badge&utm_medium=badge&utm_source=badge-parallax-by-gradient\" target=\"_blank\">\u003Cimg src=\"https:\u002F\u002Fapi.producthunt.com\u002Fwidgets\u002Fembed-image\u002Fv1\u002Ftop-post-badge.svg?post_id=1030922&theme=light&period=daily&t=1761986433128\" alt=\"Parallax by Gradient - Host LLMs across devices sharing GPU to make your AI go brrr | Product Hunt\" style=\"width: 250px; height: 54px;\" width=\"250\" height=\"54\" \u002F>\u003C\u002Fa>\n\n\u003C\u002Fdiv>\n\n| [**Gradient**](https:\u002F\u002Fgradient.network)\n| [**Blog**](https:\u002F\u002Fgradient.network\u002Fblog\u002Fparallax-the-sovereign-ai-os)\n| [**X(Twitter)(Gradient)**](https:\u002F\u002Fx.com\u002FGradient_HQ)\n| [**X(Twitter)(Parallax)**](https:\u002F\u002Fx.com\u002FtryParallax)\n| [**Discord**](https:\u002F\u002Fdiscord.gg\u002Fparallaxai)\n| [**Arxiv**](https:\u002F\u002Farxiv.org\u002Fpdf\u002F2509.26182v1)\n\n## News\n- [2026\u002F2] 🦞 Parallax now supports OpenClaw integration! See [Docs](.\u002Fdocs\u002Fuser_guide\u002Fwork_with_openclaw.md)\n- [2025\u002F10] 🔥 Parallax won #1 Product of The Day on Product Hunt!\n- [2025\u002F10] 🔥 Parallax version 0.0.1 has been released!\n\n## About\nA fully decentralized inference engine developed by [Gradient](https:\u002F\u002Fgradient.network). Parallax lets you build your own AI cluster for model inference onto a set of distributed nodes despite their varying configuration and physical location. Its core features include:\n\n- **Host local LLM on personal devices**\n- **Cross-platform support**\n- **Pipeline parallel model sharding**\n- **Paged KV cache management & continuous batching for Mac**\n- **Dynamic request scheduling and routing for high performance**\n\nThe backend architecture:\n\n* P2P communication powered by [Lattica](https:\u002F\u002Fgithub.com\u002FGradientHQ\u002Flattica)\n* GPU backend powered by [SGLang](https:\u002F\u002Fgithub.com\u002Fsgl-project\u002Fsglang) and [vLLM](https:\u002F\u002Fgithub.com\u002Fvllm-project\u002Fvllm)\n* MAC backend powered by [MLX LM](https:\u002F\u002Fgithub.com\u002Fml-explore\u002Fmlx-lm)\n\n## User Guide\n\n- [Installation](.\u002Fdocs\u002Fuser_guide\u002Finstall.md)\n- [Getting Started](.\u002Fdocs\u002Fuser_guide\u002Fquick_start.md)\n- [Working with OpenClaw 🦞](.\u002Fdocs\u002Fuser_guide\u002Fwork_with_openclaw.md)\n\n## Contributing\n\nWe warmly welcome contributions of all kinds! For guidelines on how to get involved, please refer to our [Contributing Guide](.\u002Fdocs\u002FCONTRIBUTING.md).\n\n## Supported Models\n\n|              | Provider     | HuggingFace Collection  |  Blog  | Description |\n|:-------------|:-------------|:----------------------------:|:----------------------------:|:----------------------------|\n|DeepSeek      | Deepseek     | [DeepSeek-V3.2](https:\u002F\u002Fhuggingface.co\u002Fdeepseek-ai\u002FDeepSeek-V3.2)\u003Cbr>[DeepSeek-R1](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fdeepseek-ai\u002Fdeepseek-r1) \u003Cbr>| [Deep Seek AI Launches Revolutionary Language Model](https:\u002F\u002Fdeepseek.ai\u002Fblog\u002Fdeepseek-v32) | Deep Seek AI is proud to announce the launch of our latest language model, setting new standards in natural language processing and understanding. This breakthrough represents a significant step forward in AI technology, offering unprecedented capabilities in text generation, comprehension, and analysis. |\n|MiniMax-M2    | MiniMax AI  | [MiniMax-M2](https:\u002F\u002Fhuggingface.co\u002FMiniMaxAI\u002FMiniMax-M2)\u003Cbr>[MiniMax-M2.1](https:\u002F\u002Fhuggingface.co\u002FMiniMaxAI\u002FMiniMax-M2.1) | [MiniMax M2.1: Significantly Enhanced Multi-Language Programming](https:\u002F\u002Fwww.minimax.io\u002Fnews\u002Fminimax-m21) | MiniMax-M2.1 is an enhanced sparse MoE model (230B parameters, 10B active) built for advanced coding and agentic workflows. It offers state-of-the-art intelligence, delivering efficient, reliable tool use and strong multi-step reasoning. |\n|GLM           | Z AI | [GLM-4.7](https:\u002F\u002Fhuggingface.co\u002Fzai-org\u002FGLM-4.7) \u003Cbr>[GLM-4.7-Flash](https:\u002F\u002Fhuggingface.co\u002Fzai-org\u002FGLM-4.7-Flash) | [GLM-4.7: Advancing the Coding Capability](https:\u002F\u002Fz.ai\u002Fblog\u002Fglm-4.7) | \"GLM\" is an advanced large language model series from Z AI, including GLM-4.6 and GLM-4.7. These models feature long-context support, strong coding and reasoning performance, enhanced tool-use and agent integration, and competitive results across leading open-source benchmarks. |\n|Kimi-K2       | Moonshot AI  | [Kimi-K2](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fmoonshotai\u002Fkimi-k2-6871243b990f2af5ba60617d) | [Kimi K2: Open Agentic Intelligence](https:\u002F\u002Fmoonshotai.github.io\u002FKimi-K2\u002F) | \"Kimi-K2\" is Moonshot AI's Kimi-K2 model family, including Kimi-K2-Base, Kimi-K2-Instruct and Kimi-K2-Thinking. Kimi K2 Thinking is a state-of-the-art open-source agentic model designed for deep, step-by-step reasoning and dynamic tool use. It features native INT4 quantization and a 256k context window for fast, memory-efficient inference. Uniquely stable in long-horizon tasks, Kimi K2 enables reliable autonomous workflows with consistent performance across hundreds of tool calls.\n|Qwen          | Qwen         | [Qwen3-Next](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002FQwen\u002Fqwen3-next-68c25fd6838e585db8eeea9d) \u003Cbr>[Qwen3](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002FQwen\u002Fqwen3-67dd247413f0e2e4f653967f) \u003Cbr>[Qwen2.5](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002FQwen\u002Fqwen25-66e81a666513e518adb90d9e)| [Qwen3-Next: Towards Ultimate Training & Inference Efficiency](https:\u002F\u002Fqwen.ai\u002Fblog?id=4074cca80393150c248e508aa62983f9cb7d27cd&from=research.latest-advancements-list) | The Qwen series is a family of large language models developed by Alibaba's Qwen team. It includes multiple generations such as Qwen2.5, Qwen3, and Qwen3-Next, which improve upon model architecture, efficiency, and capabilities. The models are available in various sizes and instruction-tuned versions, with support for cutting-edge features like long context and quantization. Suitable for a wide range of language tasks and open-source use cases. |\n|gpt-oss       | OpenAI       | [gpt-oss](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fopenai\u002Fgpt-oss-68911959590a1634ba11c7a4) \u003Cbr>[gpt-oss-safeguard](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fopenai\u002Fgpt-oss-safeguard) | [Introducing gpt-oss-safeguard](https:\u002F\u002Fopenai.com\u002Findex\u002Fintroducing-gpt-oss-safeguard\u002F) | gpt-oss are OpenAI’s open-weight GPT models (20B & 120B). The gpt-oss-safeguard variants are reasoning-based safety classification models: developers provide their own policy at inference, and the model uses chain-of-thought to classify content and explain its reasoning. This allows flexible, policy-driven moderation in complex or evolving domains, with open weights under Apache 2.0. |\n|Meta Llama 3  | Meta         | [Meta Llama 3](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fmeta-llama\u002Fmeta-llama-3-66214712577ca38149ebb2b6) \u003Cbr>[Llama 3.1](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fmeta-llama\u002Fllama-31-669fc079a0c406a149a5738f) \u003Cbr>[Llama 3.2](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fmeta-llama\u002Fllama-32-66f448ffc8c32f949b04c8cf) \u003Cbr>[Llama 3.3](https:\u002F\u002Fhuggingface.co\u002Fcollections\u002Fmeta-llama\u002Fllama-33-67531d5c405ec5d08a852000) | [Introducing Meta Llama 3: The most capable openly available LLM to date](https:\u002F\u002Fai.meta.com\u002Fblog\u002Fmeta-llama-3\u002F) | \"Meta Llama 3\" is Meta's third-generation Llama model, available in sizes such as 8B and 70B parameters. Includes instruction-tuned and quantized (e.g., FP8) variants. |\n","Parallax 是一个分布式模型服务框架，旨在让用户能够在任何地方构建自己的AI集群。它支持在不同配置和地理位置的节点上进行模型推理，核心功能包括本地大语言模型托管、跨平台支持、流水线并行模型切分、针对Mac的分页KV缓存管理和连续批处理以及动态请求调度与路由以实现高性能。该项目特别适合需要灵活部署AI模型的应用场景，如企业内部的私有云环境或个人设备上的AI应用开发。基于Python编写，并采用PyTorch作为主要计算后端，Parallax 为开发者提供了一个强大且易用的工具来加速其AI项目的落地。",2,"2026-06-11 03:55:36","top_topic"]