[{"data":1,"prerenderedAt":556},["ShallowReactive",2],{"project-43975423":3},{"id":4,"name":5,"line":6,"line_source":7,"title":8,"url":9,"domain":10,"maker":11,"posted_at":12,"year":13,"points":14,"comments":15,"rank_in_year":16,"shown_in_year":17,"top":18,"tier":19,"category":20,"subcategory":21,"format":22,"keywords":23,"reception":28,"sentiment":29,"stance_mix":30,"scored_comments":35,"aspects":36,"strip":44,"post_text":48,"what":49,"said":97,"maker_in_thread":220,"shape":233,"grid":237,"attempts":513,"same_week":521,"maker_shows":40,"maker_hidden":548,"reader_version":549,"card":550},"43975423","HelixDB","Open-source vector-graph database for AI applications (Rust)","title","Show HN: HelixDB – Open-source vector-graph database for AI applications (Rust)","https:\u002F\u002Fgithub.com\u002FHelixDB\u002Fhelix-db\u002F","github.com","GeorgeCurtis",1747157198,2025,237,112,259,3837,7,"notable","data","vector graph database","library",[24,25,26,27],"vector database","graph database","rag","hybrid search","curious",0.036,{"positive":31,"neutral":32,"mixed":33,"critical":34},9,38,1,8,56,{"design":37,"performance":39,"idea":41,"execution":42,"pricing":43},{"positive":38,"critical":33},0,{"positive":33,"critical":40},2,{"positive":33,"critical":40},{"positive":33,"critical":38},{"positive":38,"critical":33},[45,46,47,33,38,46,47,46,46,33,38,46,46,46,45,46,33,46,46,46,38,38,33,46,45,33,46,46,47,46,46,46,46,46,46,38,33,46,45,33],5,3,4,"Hey HN, we want to share HelixDB (https:\u002F\u002Fgithub.com\u002FHelixDB\u002Fhelix-db\u002F), a project a college friend and I are working on. It’s a new database that natively intertwines graph and vector types, without sacrificing performance. It’s written in Rust and our initial focus is on supporting RAG. Here’s a video runthrough: https:\u002F\u002Fscreen.studio\u002Fshare\u002FszgQu3yq.\n\nWhy a hybrid? Vector databases are useful for similarity queries, while graph databases are useful for relationship queries. Each stores data in a way that’s best for its main type of query (e.g. key-value stores vs. node-and-edge tables). Howe",{"problem":50,"audience":56,"pricing":60,"motivation":66,"time_to_build":70,"stack":71,"openness":94,"stage":95,"team_size":96},{"value":51,"source":52,"evidence":53},"Apps needing both similarity and relationship queries must run two databases and sync them","maker",[54],{"id":55},"post",{"value":57,"source":52,"evidence":58},"Developers building AI apps such as RAG and graph-based retrieval",[59],{"id":55},{"value":61,"source":52,"evidence":62},"Self-hosting free under AGPL-3.0; managed service planned, pricing not set",[63],{"id":64,"by":11,"text":65},"43982049","We are open-source, so you can use and self host us for free. Our plan is to create a managed service (so long as all goes well) which shouldn't be priced any differently from…",{"value":67,"source":52,"evidence":68},"Make hybrid vector and graph retrieval faster and easier for developers, after reading RAG research",[69],{"id":55},null,[72,76,88],{"value":73,"source":52,"evidence":74},"Rust",[75],{"id":55},{"value":77,"source":78,"evidence":79},"LMDB storage engine","commenter",[80,84],{"id":81,"by":82,"text":83},"43976099","hbcondo714","Congrats! Any chance Helixdb can be run in the browser too, maybe via WASM? I'm looking for a vector db that can be pre-populated on the server and then be searched on the client…",{"id":85,"by":86,"text":87},"43985706","no1youknowz","> In-house graph-vector storage engine (to replace LMDB) Not sure if it's possible. But why not use fjall, if it is? [0] [0]: https:\u002F\u002Fgithub.com\u002Ffjall-rs\u002Ffjall\u002F",{"value":89,"source":52,"evidence":90},"Python SDK",[91],{"id":92,"by":11,"text":93},"43982162","We started as a graph database, so that's definitely the main thing we want to get right and we wan't to prioritise capturing all that functionality. We have a python SDK…","open_source","beta","team",{"praise":98,"objections":99,"requests":133,"questions":164,"alternatives":197},[],[100,111,122],{"value":101,"source":78,"evidence":102},"Bespoke query language creates lock-in and learning curve",[103,107],{"id":104,"by":105,"text":106},"43977005","mdaniel","> so much easier that it’s worth a bit of a learning curve I think you misspelled \"vendor lock in\"",{"id":108,"by":109,"text":110},"43984707","bogzz","This is very cool, and right up my alley. Hesitant to try it out because of the bespoke query language for now. I wonder if you'd like to share your thoughts on GQL becoming an…",{"value":112,"source":78,"evidence":113},"Benchmark claims were not backed by published numbers",[114,118],{"id":115,"by":116,"text":117},"43977681","Attummm","It sounds very intriguing indeed. However, the README makes some claims. Are there any benchmarks to support them? > Built for performance we're currently 1000x faster than…",{"id":119,"by":120,"text":121},"43978307","lennertjansen","how did you get it 3 OOMs faster than neo4j?",{"value":123,"source":78,"evidence":124},"Name clashes with the Helix editor and an existing Helix database",[125,129],{"id":126,"by":127,"text":128},"43976348","lillecarl","Nice \"I'll have this name\" when there's already the helix editor :)",{"id":130,"by":131,"text":132},"43977387","itishappy","I don't think `helix-editor` is even on crates.io, just placeholders. https:\u002F\u002Fgithub.com\u002Fhelix-editor\u002Fhelix\u002Fdiscussions\u002F7038 That being said, when I saw `helix-db` I was thrown…",[134,149,157],{"value":135,"source":78,"evidence":136},"Published benchmarks against other graph and vector databases",[137,141,145],{"id":138,"by":139,"text":140},"43975955","esafak","How does it compare with https:\u002F\u002Fkuzudb.com\u002F ?",{"id":142,"by":143,"text":144},"43985151","iamdanieljohns","How does it compare to SurrealDB and ChromaDB?",{"id":146,"by":147,"text":148},"43980575","wiradikusuma","\"faster than Neo4j\" How does it compare to Dgraph?",{"value":150,"source":78,"evidence":151},"Embedded mode or browser WASM support",[152,153],{"id":81,"by":82,"text":83},{"id":154,"by":155,"text":156},"43976519","huevosabio","Can I run this as an embedded DB like sqlite? Can I sidestep the DSL? I want my LLMs to generate queries and using a new language is going to make that hard or expensive.",{"value":158,"source":78,"evidence":159},"Python SDK with generated types",[160],{"id":161,"by":162,"text":163},"43978573","youdont","Looks very interesting, but I've seen these kind of multi-paradigm databases like Gel, Helix and Surreal and I'm not sure that any of them quite hit the graph spot. Does Helix…",[165,178,182],{"value":166,"source":78,"evidence":167},"How it compares to Kuzu, SurrealDB, ChromaDB, Memgraph and CozoDB",[168,169,170,174],{"id":138,"by":139,"text":140},{"id":142,"by":143,"text":144},{"id":171,"by":172,"text":173},"43984482","ckugblenu","Very interesting project. Would be curious of a comparison with memgraph. Will definitely give it to try for my knowledge graph use case.",{"id":175,"by":176,"text":177},"43980555","michaelsbradley","Can you do a compare\u002Fcontrast with CozoDB? https:\u002F\u002Fgithub.com\u002Fcozodb\u002Fcozo",{"value":179,"source":78,"evidence":180},"Whether it can run embedded like SQLite",[181],{"id":154,"by":155,"text":156},{"value":183,"source":78,"evidence":184},"Hardware and scaling requirements",[185,189,193],{"id":186,"by":187,"text":188},"43979626","srameshc","I was thinking about intertwining Vector and Graph, because I have one specific usecase that required this combination. But I am not courageos or competent enough to build such a…",{"id":190,"by":191,"text":192},"43980885","anonymousDan","What would be a typical\u002Frecommended server setup for using this for RAG? Would you typically have a separate server for the GPUs and the DB itself?",{"id":194,"by":195,"text":196},"43978700","riku_iki","How scalable is your DB in your tests? Could it be performent on graphs with 1B\u002F10B\u002F100B connections?",[198,205,209,216],{"value":199,"source":78,"evidence":200},"Neo4j",[201],{"id":202,"by":203,"text":204},"43980247","UltraSane","Neo4j supports vector indexes",{"value":206,"source":78,"evidence":207},"Kuzu",[208],{"id":138,"by":139,"text":140},{"value":210,"source":78,"evidence":211},"SurrealDB",[212],{"id":213,"by":214,"text":215},"43977294","basonjourne","why not surrealdb?",{"value":217,"source":52,"evidence":218},"Qdrant and Pinecone",[219],{"id":55},{"comments":221,"answered":222,"key":223},41,28,[224,227,230],{"id":225,"text":226},"43978242","Wow! I enjoyed reading it a lot and it was definitely inspiring for this project! Would love to talk to you about it and make sure we capture all of the pain points if you're…",{"id":228,"text":229},"43980477","Neo4j first of all is very slow for vectors, so if performance is something that matters for your user experience they definitely aren't a viable option. This is probably why…",{"id":231,"text":232},"43992766","Thanks for clearing up some of these points. Really pleased the post has reached the industry leaders, and I genuinely appreciate your response :) With regards to the…",{"commenters":234,"max_depth":18,"maker_replies":221,"first_comment_mins":235,"lifespan_hours":236},51,37,345,[238,242,244,246,250,253,255,258,260,262,264,268,269,272,274,276,278,281,284,287,288,290,293,295,298,300,304,306,307,311,313,314,318,320,322,326,329,331,332,334,339,341,343,345,347,349,354,356,359,361,364,366,368,370,374,376,378,380,382,384,386,388,392,394,395,398,401,403,407,409,414,415,418,421,425,427,429,431,433,435,437,439,441,445,447,449,451,455,457,459,461,463,465,467,469,471,474,476,478,481,483,484,486,489,491,495,497,501,503,507,509,511],[239,38,45,240,241,38],"43977985","maker\u002Fpositive\u002Fnone","rohanrao123",[225,38,38,243,11,33],"commenter\u002Fneutral\u002Fnone",[245,38,38,243,241,38],"43978364",[247,33,46,248,249,38],"43981161","project\u002Fneutral\u002Fdesign","quantike",[251,33,46,248,252,38],"43982324","xavcochran",[254,33,46,248,249,38],"43990355",[256,33,46,257,252,38],"43990552","project\u002Fneutral\u002Fperformance",[186,40,45,259,187,38],"project\u002Fpositive\u002Fidea",[202,40,33,261,203,38],"project\u002Fcritical\u002Fidea",[228,40,38,263,11,33],"project\u002Fpositive\u002Fperformance",[265,40,38,266,267,38],"43990034","commenter\u002Fneutral\u002Fperformance","mauvo59",[231,40,38,243,11,33],[270,40,38,266,271,38],"44112261","acefaceZ",[273,40,38,257,11,33],"43980412",[81,46,46,275,82,38],"project\u002Fneutral\u002Fidea",[277,46,38,275,11,33],"43976205",[279,46,46,280,252,38],"43982342","project\u002Fneutral\u002Fexecution",[282,46,46,280,283,38],"43982626","hansworst",[285,46,45,286,252,38],"43982711","project\u002Fpositive\u002Fexecution",[154,47,46,248,155,38],[289,47,38,248,11,33],"43977195",[291,45,46,248,292,38],"43978122","tmpfs",[294,45,38,275,11,33],"43978392",[138,296,46,297,139,38],6,"project\u002Fneutral\u002Fnone",[299,296,38,275,11,33],"43976039",[301,296,33,302,303,38],"43999780","project\u002Fcritical\u002Fdesign","wontonaroo",[305,296,38,248,11,33],"44001668",[161,18,46,275,162,38],[308,18,38,309,310,38],"43980119","other\u002Fneutral\u002Fnone","Onawa",[312,18,38,243,11,33],"43980517",[92,18,38,275,11,33],[315,18,38,316,317,38],"43982654","other\u002Fneutral\u002Fperformance","BlooIt",[319,18,46,257,252,38],"43987261",[321,18,38,316,317,38],"44029426",[323,18,38,324,325,38],"44068342","other\u002Fneutral\u002Fpricing","mbrinman",[327,34,46,280,328,38],"43987231","iannacl",[330,34,46,280,252,38],"43987493",[115,31,46,257,116,38],[333,31,38,257,11,33],"43977722",[335,336,45,337,338,38],"43980055",10,"project\u002Fpositive\u002Fnone","sitkack",[340,336,46,280,252,38],"43982060",[171,342,45,337,172,38],11,[344,342,38,297,11,33],"43987144",[142,346,46,297,143,38],12,[348,346,38,257,11,33],"43989616",[350,351,33,352,353,38],"43978692",13,"project\u002Fcritical\u002Fperformance","dietr1ch",[355,351,38,263,11,33],"43982006",[357,351,46,297,358,38],"44036990","grounder",[360,351,38,337,11,33],"44077489",[108,362,46,363,109,38],14,"project\u002Fmixed\u002Fdesign",[365,362,38,275,11,33],"43989598",[85,367,46,280,86,38],15,[369,367,38,257,11,33],"43987130",[371,372,46,275,373,38],"43976157",16,"J_Shelby_J",[375,372,38,275,11,33],"43976299",[126,377,38,309,127,38],17,[379,377,38,309,11,33],"43976458",[381,377,38,309,127,38],"43976524",[383,377,38,297,11,33],"43976975",[385,377,38,243,127,38],"43977010",[387,377,38,243,11,33],"43977602",[389,377,33,390,391,38],"43977505","project\u002Fcritical\u002Fnone","tavianator",[393,377,38,243,11,33],"43977612",[130,377,33,390,131,38],[396,377,38,243,397,38],"43976871","bbatsell",[399,377,46,297,400,38],"43976731","cormullion",[402,377,38,243,11,33],"43977160",[404,405,45,337,406,38],"43976055",18,"SchwKatze",[408,405,38,243,11,33],"43976126",[410,411,46,412,413,38],"43980734",19,"project\u002Fneutral\u002Fpricing","rationably",[64,411,38,412,11,33],[416,411,33,417,413,38],"43995878","project\u002Fcritical\u002Fpricing",[419,411,38,420,11,33],"43997614","commenter\u002Fneutral\u002Fpricing",[422,423,46,257,424,38],"43976392",20,"javierluraschi",[426,423,38,257,11,33],"43976529",[428,423,38,309,252,38],"43982793",[190,430,46,297,191,38],21,[432,430,46,297,252,38],"43982204",[175,434,46,275,176,38],22,[436,434,45,263,252,38],"43982297",[194,438,46,257,195,38],23,[440,438,38,257,11,33],"43980790",[442,443,46,257,444,38],"43987628",24,"brene",[446,443,38,257,11,33],"43987895",[146,448,46,257,147,38],25,[450,448,38,257,11,33],"43980911",[452,453,46,275,454,38],"43976983",26,"elpalek",[456,453,38,275,11,33],"43977204",[458,453,46,275,454,38],"43977253",[460,453,38,243,11,33],"43977599",[462,453,46,257,454,38],"43978518",[464,453,46,275,252,38],"43982353",[466,453,38,243,252,38],"43983962",[213,468,46,275,214,38],27,[470,468,38,257,11,33],"43978140",[472,468,46,257,473,38],"43984050","datastorydesign",[475,468,38,243,11,33],"43990931",[477,468,38,309,195,38],"44005240",[479,468,38,243,480,38],"43990695","mattturck",[482,468,38,243,11,33],"43990941",[119,222,46,257,120,38],[485,222,38,257,11,33],"43978430",[487,222,33,352,488,38],"43978884","xpe",[490,222,38,257,11,33],"43982135",[492,493,46,275,494,38],"43978123",29,"raufakdemir",[496,493,38,297,11,33],"43978163",[498,499,45,337,500,38],"43982933",30,"lleymrl651",[502,499,45,337,252,38],"43984035",[504,505,46,275,506,38],"43975838",31,"sync",[508,505,38,275,11,33],"43975920",[104,510,33,261,105,38],32,[512,510,38,243,11,33],"43977638",[514,517,520],{"id":515,"t":516,"points":47,"comments":47},"43155338",1740365043,{"id":518,"t":519,"points":448,"comments":296},"43659999",1744415678,{"id":4,"t":12,"points":14,"comments":15},[522,527,531,536,541,545],{"id":523,"name":524,"points":525,"category":526},"43935394","Hyvector",331,"dev_tools",{"id":528,"name":529,"points":16,"category":530},"43928118","qtap","infrastructure",{"id":532,"name":533,"points":534,"category":535},"43988381","Muscle-Mem",226,"ai",{"id":537,"name":538,"points":539,"category":540},"43936097","Aberdeen",224,"languages",{"id":542,"name":543,"points":544,"category":535},"43964201","Airweave",176,{"id":546,"name":547,"points":544,"category":535},"43988533","Semantic Calculator",false,"v2",{"id":551,"name":5,"line":6,"ai":548,"cat":20,"year":13,"points":14,"comments":15,"rec":28,"s":552,"top":18,"kw":553,"maker":11,"domain":10,"tier":19,"attempts":46,"strip":554,"url":9,"mix":555,"praise":-1,"objection":101,"sub":21},43975423,0.04,[24,25,26,27],"5341034331033353133300135133433333301351",[31,32,33,34],1794036674741]