[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill-qdrant-qdrant-minimize-latency":3,"mdc--o1wngf-key":48,"related-repo-qdrant-qdrant-minimize-latency":268,"related-org-qdrant-qdrant-minimize-latency":366},{"slug":4,"name":4,"fn":5,"description":6,"org":7,"tags":11,"stars":23,"repoUrl":24,"updatedAt":25,"license":26,"forks":27,"topics":28,"repo":43,"sourceUrl":46,"mdContent":47},"qdrant-minimize-latency","optimize Qdrant query latency","Guides Qdrant query latency optimization. Use when someone asks 'search is slow', 'how to reduce latency', 'p99 is too high', 'tail latency', 'single query too slow', 'how to make search faster', or 'latency spikes'.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},"qdrant","Qdrant","https:\u002F\u002Fpexgzepcugksgbtrxkhf.supabase.co\u002Fstorage\u002Fv1\u002Fobject\u002Fpublic\u002Forg-logos\u002Fqdrant.png",[12,16,17,20],{"name":13,"slug":14,"type":15},"Performance","performance","tag",{"name":9,"slug":8,"type":15},{"name":18,"slug":19,"type":15},"Database","database",{"name":21,"slug":22,"type":15},"Debugging","debugging",197,"https:\u002F\u002Fgithub.com\u002Fqdrant\u002Fskills","2026-07-16T05:59:17.325839",null,23,[29,30,31,32,33,34,35,36,37,14,8,38,39,40,41,42],"agent-skills","ai-agents","claude-code","codex","cursor","embeddings","hybrid-search","monitoring","multitenancy","quantization","scaling","search-quality","vector-database","vector-search",{"repoUrl":24,"stars":23,"forks":27,"topics":44,"description":45},[29,30,31,32,33,34,35,36,37,14,8,38,39,40,41,42],"Agent skills for Qdrant vector search: scaling, performance optimization, search quality, monitoring, deployment, model migration, version upgrades, and SDK usage across Python, TypeScript, Rust, Go, .NET, Java","https:\u002F\u002Fgithub.com\u002Fqdrant\u002Fskills\u002Ftree\u002FHEAD\u002Fskills\u002Fqdrant-scaling\u002Fminimize-latency","---\nname: qdrant-minimize-latency\ndescription: \"Guides Qdrant query latency optimization. Use when someone asks 'search is slow', 'how to reduce latency', 'p99 is too high', 'tail latency', 'single query too slow', 'how to make search faster', or 'latency spikes'.\"\n---\n\n# Scaling for Query Latency\n\nLatency of a single query is determined by the slowest component in the query execution path. It is sometimes correlated with throughput, but not always — throughput and latency are opposite tuning directions.\n\nLow latency optimization is aimed at utilising maximum resource saturation for a single query, while throughput optimization is aimed at minimizing per-query resource usage to allow more parallel queries.\n\n## Performance Tuning for Lower Latency\n\n- Increase segment count to match CPU cores (`default_segment_number: 16`) [Minimizing latency](https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fops-optimization\u002Foptimize\u002F?s=minimizing-latency)\n- Keep quantized vectors and HNSW in RAM (`always_ram=true`)\n- Reduce `hnsw_ef` at query time (trade recall for speed) [Search params](https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fops-optimization\u002Foptimize\u002F?s=fine-tuning-search-parameters)\n- Use local NVMe, avoid network-attached storage\n\n## Memory Pressure and Latency\n\nRAM is the most critical resource for latency. If working set exceeds available RAM, OS cache eviction causes severe, sustained latency degradation.\n\n- Vertical scale RAM first. Critical if working set >80%.\n- Use quantization: scalar (4x reduction) or binary (16x reduction) [Quantization](https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fmanage-data\u002Fquantization\u002F)\n- Move payload indexes to disk if filtering is infrequent [On-disk payload index](https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fmanage-data\u002Findexing\u002F?s=on-disk-payload-index)\n- Set `optimizer_cpu_budget` to limit background optimization CPUs\n- Schedule indexing: set high `indexing_threshold` during peak hours\n\n\n## Vertical Scaling for Latency\n\nMore RAM and faster CPU directly reduce latency. See [Vertical Scaling](..\u002Fscaling-data-volume\u002Fvertical-scaling\u002FSKILL.md) for node sizing guidelines.\n\n\n## What NOT to Do\n\n- Do not expect to optimize latency and throughput simultaneously on the same node\n- Do not use few large segments for latency-sensitive workloads (each segment takes longer to search)\n- Do not run at >90% RAM (cache eviction causes severe latency degradation that can last days)\n- Do not ignore optimizer status during performance debugging\n- Do not scale down RAM without load testing (cache eviction causes days-long latency incidents)\n",{"data":49,"body":50},{"name":4,"description":6},{"type":51,"children":52},"root",[53,62,68,73,80,146,152,157,215,221,234,240],{"type":54,"tag":55,"props":56,"children":58},"element","h1",{"id":57},"scaling-for-query-latency",[59],{"type":60,"value":61},"text","Scaling for Query Latency",{"type":54,"tag":63,"props":64,"children":65},"p",{},[66],{"type":60,"value":67},"Latency of a single query is determined by the slowest component in the query execution path. It is sometimes correlated with throughput, but not always — throughput and latency are opposite tuning directions.",{"type":54,"tag":63,"props":69,"children":70},{},[71],{"type":60,"value":72},"Low latency optimization is aimed at utilising maximum resource saturation for a single query, while throughput optimization is aimed at minimizing per-query resource usage to allow more parallel queries.",{"type":54,"tag":74,"props":75,"children":77},"h2",{"id":76},"performance-tuning-for-lower-latency",[78],{"type":60,"value":79},"Performance Tuning for Lower Latency",{"type":54,"tag":81,"props":82,"children":83},"ul",{},[84,108,121,141],{"type":54,"tag":85,"props":86,"children":87},"li",{},[88,90,97,99],{"type":60,"value":89},"Increase segment count to match CPU cores (",{"type":54,"tag":91,"props":92,"children":94},"code",{"className":93},[],[95],{"type":60,"value":96},"default_segment_number: 16",{"type":60,"value":98},") ",{"type":54,"tag":100,"props":101,"children":105},"a",{"href":102,"rel":103},"https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fops-optimization\u002Foptimize\u002F?s=minimizing-latency",[104],"nofollow",[106],{"type":60,"value":107},"Minimizing latency",{"type":54,"tag":85,"props":109,"children":110},{},[111,113,119],{"type":60,"value":112},"Keep quantized vectors and HNSW in RAM (",{"type":54,"tag":91,"props":114,"children":116},{"className":115},[],[117],{"type":60,"value":118},"always_ram=true",{"type":60,"value":120},")",{"type":54,"tag":85,"props":122,"children":123},{},[124,126,132,134],{"type":60,"value":125},"Reduce ",{"type":54,"tag":91,"props":127,"children":129},{"className":128},[],[130],{"type":60,"value":131},"hnsw_ef",{"type":60,"value":133}," at query time (trade recall for speed) ",{"type":54,"tag":100,"props":135,"children":138},{"href":136,"rel":137},"https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fops-optimization\u002Foptimize\u002F?s=fine-tuning-search-parameters",[104],[139],{"type":60,"value":140},"Search params",{"type":54,"tag":85,"props":142,"children":143},{},[144],{"type":60,"value":145},"Use local NVMe, avoid network-attached storage",{"type":54,"tag":74,"props":147,"children":149},{"id":148},"memory-pressure-and-latency",[150],{"type":60,"value":151},"Memory Pressure and Latency",{"type":54,"tag":63,"props":153,"children":154},{},[155],{"type":60,"value":156},"RAM is the most critical resource for latency. If working set exceeds available RAM, OS cache eviction causes severe, sustained latency degradation.",{"type":54,"tag":81,"props":158,"children":159},{},[160,165,177,189,202],{"type":54,"tag":85,"props":161,"children":162},{},[163],{"type":60,"value":164},"Vertical scale RAM first. Critical if working set >80%.",{"type":54,"tag":85,"props":166,"children":167},{},[168,170],{"type":60,"value":169},"Use quantization: scalar (4x reduction) or binary (16x reduction) ",{"type":54,"tag":100,"props":171,"children":174},{"href":172,"rel":173},"https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fmanage-data\u002Fquantization\u002F",[104],[175],{"type":60,"value":176},"Quantization",{"type":54,"tag":85,"props":178,"children":179},{},[180,182],{"type":60,"value":181},"Move payload indexes to disk if filtering is infrequent ",{"type":54,"tag":100,"props":183,"children":186},{"href":184,"rel":185},"https:\u002F\u002Fskills.qdrant.tech\u002Fmd\u002Fdocumentation\u002Fmanage-data\u002Findexing\u002F?s=on-disk-payload-index",[104],[187],{"type":60,"value":188},"On-disk payload index",{"type":54,"tag":85,"props":190,"children":191},{},[192,194,200],{"type":60,"value":193},"Set ",{"type":54,"tag":91,"props":195,"children":197},{"className":196},[],[198],{"type":60,"value":199},"optimizer_cpu_budget",{"type":60,"value":201}," to limit background optimization CPUs",{"type":54,"tag":85,"props":203,"children":204},{},[205,207,213],{"type":60,"value":206},"Schedule indexing: set high ",{"type":54,"tag":91,"props":208,"children":210},{"className":209},[],[211],{"type":60,"value":212},"indexing_threshold",{"type":60,"value":214}," during peak hours",{"type":54,"tag":74,"props":216,"children":218},{"id":217},"vertical-scaling-for-latency",[219],{"type":60,"value":220},"Vertical Scaling for Latency",{"type":54,"tag":63,"props":222,"children":223},{},[224,226,232],{"type":60,"value":225},"More RAM and faster CPU directly reduce latency. See ",{"type":54,"tag":100,"props":227,"children":229},{"href":228},"..\u002Fscaling-data-volume\u002Fvertical-scaling\u002FSKILL.md",[230],{"type":60,"value":231},"Vertical Scaling",{"type":60,"value":233}," for node sizing guidelines.",{"type":54,"tag":74,"props":235,"children":237},{"id":236},"what-not-to-do",[238],{"type":60,"value":239},"What NOT to Do",{"type":54,"tag":81,"props":241,"children":242},{},[243,248,253,258,263],{"type":54,"tag":85,"props":244,"children":245},{},[246],{"type":60,"value":247},"Do not expect to optimize latency and throughput simultaneously on the same node",{"type":54,"tag":85,"props":249,"children":250},{},[251],{"type":60,"value":252},"Do not use few large segments for latency-sensitive workloads (each segment takes longer to search)",{"type":54,"tag":85,"props":254,"children":255},{},[256],{"type":60,"value":257},"Do not run at >90% RAM (cache eviction causes severe latency degradation that can last days)",{"type":54,"tag":85,"props":259,"children":260},{},[261],{"type":60,"value":262},"Do not ignore optimizer status during performance debugging",{"type":54,"tag":85,"props":264,"children":265},{},[266],{"type":60,"value":267},"Do not scale down RAM without load testing (cache eviction causes days-long latency incidents)",{"items":269,"total":365},[270,283,298,313,328,342,355],{"slug":271,"name":271,"fn":272,"description":273,"org":274,"tags":275,"stars":23,"repoUrl":24,"updatedAt":282},"qdrant-advisor","diagnose and troubleshoot Qdrant deployments","Diagnose, troubleshoot, and advise on any Qdrant deployment by loading the latest official Qdrant skills live from skills.qdrant.tech. Use this whenever someone raises a Qdrant problem or question — slow or degraded search, high or growing memory \u002F OOM crashes, optimizer stuck or slow, indexing slowness, scaling and sharding decisions (node count, QPS, latency, multitenancy, vertical vs horizontal), poor or irrelevant search results, hybrid search and reranking, embedding-model migration, version upgrades and compatibility, monitoring and observability (Prometheus, Grafana, health checks, \u002Fmetrics, \u002Ftelemetry), deployment choices (local, Docker, self-hosted, Qdrant Cloud, embedded), or client-SDK questions (Python, TypeScript, Rust, Go, .NET, Java). Trigger especially when the context is clearly a Qdrant cluster, collection, or vector-search deployment. Always prefer this skill over answering from memory: it pulls current, authoritative guidance and only the relevant context.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[276,277,278,281],{"name":18,"slug":19,"type":15},{"name":21,"slug":22,"type":15},{"name":279,"slug":280,"type":15},"Operations","operations",{"name":9,"slug":8,"type":15},"2026-07-16T06:02:56.675293",{"slug":284,"name":284,"fn":285,"description":286,"org":287,"tags":288,"stars":23,"repoUrl":24,"updatedAt":297},"qdrant-clients-sdk","integrate Qdrant client SDKs","Qdrant provides client SDKs for various programming languages, allowing easy integration with Qdrant deployments.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[289,292,293,294],{"name":290,"slug":291,"type":15},"API Development","api-development",{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":295,"slug":296,"type":15},"SDK","sdk","2026-07-16T05:59:57.799789",{"slug":299,"name":299,"fn":300,"description":301,"org":302,"tags":303,"stars":23,"repoUrl":24,"updatedAt":312},"qdrant-deployment-options","select Qdrant deployment options","Guides Qdrant deployment selection. Use when someone asks 'how to deploy Qdrant', 'Docker vs Cloud', 'local mode', 'embedded Qdrant', 'Qdrant EDGE', 'which deployment option', 'self-hosted vs cloud', or 'need lowest latency deployment'. Also use when choosing between deployment types for a new project.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[304,305,308,311],{"name":18,"slug":19,"type":15},{"name":306,"slug":307,"type":15},"Deployment","deployment",{"name":309,"slug":310,"type":15},"Infrastructure","infrastructure",{"name":9,"slug":8,"type":15},"2026-07-19T05:38:33.324265",{"slug":314,"name":314,"fn":315,"description":316,"org":317,"tags":318,"stars":23,"repoUrl":24,"updatedAt":327},"qdrant-edge","build applications with Qdrant Edge","Guides building on Qdrant Edge, the embedded in-process shard. Use when someone asks 'how to sync Edge with the server', 'keep a local shard in sync with Qdrant Cloud', 'BM25 or keyword search on Edge', 'hybrid search on Edge', 'embeddings on device', 'Edge snapshots', 'apply a partial snapshot', 'why is my Edge search empty after inserts', or is writing custom sync, BM25, or fusion code against qdrant-edge. Also use when deciding what Edge ships built-in versus what you must implement.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[319,320,323,324],{"name":18,"slug":19,"type":15},{"name":321,"slug":322,"type":15},"Edge","edge",{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},"Search","search","2026-07-16T06:01:19.179105",{"slug":329,"name":329,"fn":330,"description":331,"org":332,"tags":333,"stars":23,"repoUrl":24,"updatedAt":341},"qdrant-horizontal-scaling","guide Qdrant horizontal scaling decisions","Diagnoses and guides Qdrant horizontal scaling decisions. Use when someone asks 'vertical or horizontal?', 'how many nodes?', 'how many shards?', 'how to add nodes', 'resharding', 'data doesn't fit', or 'need more capacity'. Also use when data growth outpaces current deployment.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[334,337,338,339],{"name":335,"slug":336,"type":15},"Architecture","architecture",{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":340,"slug":39,"type":15},"Scaling","2026-07-19T05:38:32.348551",{"slug":343,"name":343,"fn":344,"description":345,"org":346,"tags":347,"stars":23,"repoUrl":24,"updatedAt":354},"qdrant-hybrid-search","implement hybrid search in Qdrant","Explains hybrid search in Qdrant. Use when someone asks 'how do I setup hybrid search?', 'how to combine keyword and semantic search?', 'sparse plus dense vectors?', 'missing keyword matches', 'how to combine results from multiple searches?' and 'combining multiple representations'",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[348,351,352,353],{"name":349,"slug":350,"type":15},"AI Infrastructure","ai-infrastructure",{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},"2026-07-16T06:00:12.328122",{"slug":356,"name":356,"fn":357,"description":358,"org":359,"tags":360,"stars":23,"repoUrl":24,"updatedAt":364},"qdrant-hybrid-search-combining","combine hybrid search scores in Qdrant","Fusing scores from multiple searches into a single ranked result (RRF, DBSF, custom fusion). Use when someone asks 'RRF or DBSF?', 'how to combine sparse and dense', 'how to combine scores from multiple searches?', 'custom fusion', or 'fusion is not producing good results'",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[361,362,363],{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},"2026-07-16T06:01:49.401142",30,{"items":367,"total":365},[368,375,382,389,396,403,410,416,427,438,448,455],{"slug":271,"name":271,"fn":272,"description":273,"org":369,"tags":370,"stars":23,"repoUrl":24,"updatedAt":282},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[371,372,373,374],{"name":18,"slug":19,"type":15},{"name":21,"slug":22,"type":15},{"name":279,"slug":280,"type":15},{"name":9,"slug":8,"type":15},{"slug":284,"name":284,"fn":285,"description":286,"org":376,"tags":377,"stars":23,"repoUrl":24,"updatedAt":297},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[378,379,380,381],{"name":290,"slug":291,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":295,"slug":296,"type":15},{"slug":299,"name":299,"fn":300,"description":301,"org":383,"tags":384,"stars":23,"repoUrl":24,"updatedAt":312},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[385,386,387,388],{"name":18,"slug":19,"type":15},{"name":306,"slug":307,"type":15},{"name":309,"slug":310,"type":15},{"name":9,"slug":8,"type":15},{"slug":314,"name":314,"fn":315,"description":316,"org":390,"tags":391,"stars":23,"repoUrl":24,"updatedAt":327},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[392,393,394,395],{"name":18,"slug":19,"type":15},{"name":321,"slug":322,"type":15},{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},{"slug":329,"name":329,"fn":330,"description":331,"org":397,"tags":398,"stars":23,"repoUrl":24,"updatedAt":341},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[399,400,401,402],{"name":335,"slug":336,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":340,"slug":39,"type":15},{"slug":343,"name":343,"fn":344,"description":345,"org":404,"tags":405,"stars":23,"repoUrl":24,"updatedAt":354},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[406,407,408,409],{"name":349,"slug":350,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},{"slug":356,"name":356,"fn":357,"description":358,"org":411,"tags":412,"stars":23,"repoUrl":24,"updatedAt":364},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[413,414,415],{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},{"slug":417,"name":417,"fn":418,"description":419,"org":420,"tags":421,"stars":23,"repoUrl":24,"updatedAt":426},"qdrant-hybrid-search-prefetches","configure hybrid search with prefetch queries","Constructing prefetch queries for hybrid retrieval, including sparse\u002Fdense and multi-field setups, and choosing a sparse embedding model. Use when someone asks 'dense and sparse in one search?', 'how to combine multiple fields for retrieval?', 'payloads or sparse vectors for lexical?', 'which sparse embedding model to use?', or 'BM25 vs SPLADE?'",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[422,423,424,425],{"name":290,"slug":291,"type":15},{"name":18,"slug":19,"type":15},{"name":9,"slug":8,"type":15},{"name":325,"slug":326,"type":15},"2026-07-16T06:03:02.305907",{"slug":428,"name":428,"fn":429,"description":430,"org":431,"tags":432,"stars":23,"repoUrl":24,"updatedAt":437},"qdrant-indexing-performance-optimization","optimize Qdrant indexing and data ingestion","Diagnoses and fixes slow Qdrant indexing and data ingestion. Use when someone reports 'uploads are slow', 'indexing takes forever', 'optimizer is stuck', 'HNSW build time too long', or 'data uploaded but search is bad'. Also use when optimizer status shows errors, segments won't merge, or indexing threshold questions arise.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[433,434,435,436],{"name":18,"slug":19,"type":15},{"name":21,"slug":22,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-07-19T05:38:36.321999",{"slug":439,"name":439,"fn":440,"description":441,"org":442,"tags":443,"stars":23,"repoUrl":24,"updatedAt":447},"qdrant-memory-usage-optimization","optimize Qdrant memory usage","Diagnoses and reduces Qdrant memory usage. Use when someone reports 'memory too high', 'RAM keeps growing', 'node crashed', 'out of memory', 'memory leak', or asks 'why is memory usage so high?', 'how to reduce RAM?'. Also use when memory doesn't match calculations, quantization didn't help, or nodes crash during recovery.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[444,445,446],{"name":21,"slug":22,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-07-16T06:00:15.357243",{"slug":4,"name":4,"fn":5,"description":6,"org":449,"tags":450,"stars":23,"repoUrl":24,"updatedAt":25},{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[451,452,453,454],{"name":18,"slug":19,"type":15},{"name":21,"slug":22,"type":15},{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},{"slug":456,"name":456,"fn":457,"description":458,"org":459,"tags":460,"stars":23,"repoUrl":24,"updatedAt":467},"qdrant-model-migration","migrate embedding models in Qdrant","Guides embedding model migration in Qdrant without downtime. Use when someone asks 'how to switch embedding models', 'how to migrate vectors', 'how to update to a new model', 'zero-downtime model change', 'how to re-embed my data', or 'can I use two models at once'. Also use when upgrading model dimensions, switching providers, or A\u002FB testing models.",{"slug":8,"name":9,"logoUrl":10,"githubOrg":8},[461,462,465,466],{"name":18,"slug":19,"type":15},{"name":463,"slug":464,"type":15},"Migration","migration",{"name":13,"slug":14,"type":15},{"name":9,"slug":8,"type":15},"2026-07-16T06:00:48.841055"]