Qdrant
AI & LLMsQdrant - High-performance, massive-scale Vector Database and Vector Search Engine for the next generation of AI. Also available in the cloud https://cloud.qdrant.io/
Release History
v1.17.020 fixes13 featuresThis release introduces significant stability and performance improvements, focusing on distributed cluster operations, snapshot recovery, and reduced locking overhead. Key new features include Relevance Feedback and enhanced monitoring APIs.
v1.16.310 fixesThis release focuses on stability and data integrity, fixing several critical bugs related to WAL transfers, data flushing under error conditions, and collection name handling. Improvements were also made to respect various timeouts during operations.
v1.16.29 fixes1 featureThis release focuses heavily on stability and data integrity, fixing numerous critical bugs related to WAL, consensus, storage flushing, and resource cleanup. Improvements include better request timeout handling and adding a user agent to outgoing HTTP requests.
v1.16.17 fixes7 featuresThis release focuses heavily on performance improvements, including faster batch queries and active storage migration to Gridstore, alongside numerous stability fixes for Raft, WAL handling, and shard transfers.
v1.16.010 fixes12 featuresThis release introduces significant performance improvements through AVX512 optimizations and HNSW inline storage, alongside major features like tiered multitenancy support and ACORN search algorithm. Numerous bug fixes address stability issues related to cancellation, snapshots, and cluster operations.
v1.15.516 fixes3 featuresThis release focuses heavily on stability and performance, introducing numerous bug fixes for deadlocks and data races, alongside improvements in API validation and resource management to prevent OOM errors.
v1.15.43 fixes5 featuresThis release focuses on improving Docker image efficiency and security, enhancing disk space measurement reliability, and fixing several critical bugs related to index loading and segment proxying.
v1.15.31 fixThis release focuses on performance improvements, including optimized dot product calculation on AVX systems, and fixes an issue where local BM25 was not behaving as expected.
v1.15.29 fixes4 featuresThis release introduces BM25 inference support and performance improvements for mutable map indexes. Several consistency issues related to index storage and point deletion have been resolved.
v1.15.14 fixes1 featureThis release improves IO performance during indexing and fixes several critical bugs, including issues with point shard routing in 1.15 and UUID index storage.
v1.15.011 fixes8 featuresThis release introduces significant enhancements to Full-Text indexing, including phrase matching and stop words support, alongside new binary quantization options and performance improvements across the board. Several older configuration parameters and APIs are deprecated in favor of more modern and consistent alternatives.
v1.14.110 fixes1 featureThis release focuses heavily on performance improvements across WAL transfer, GPU indexing, and payload index loading by replacing RocksDB with mmaps. Several bugs related to strict mode enforcement, upsert behavior, and index consistency have also been resolved.
v1.14.04 fixes3 featuresThis release introduces server-side score boosting and the new `sum_scores` recommendation strategy. Improvements focus on incremental HNSW building and better disk cache eviction, alongside several stability fixes for shard recovery and data consistency.
v1.13.62 fixesThis release focuses heavily on performance improvements in the query API by optimizing vector/payload handling and significantly speeding up resharding transfers. Several bugs related to resharding estimation and ordering in queries were also fixed.
v1.13.516 fixes2 featuresThis release focuses heavily on stability and performance improvements, particularly around cluster operations like resharding and snapshotting, alongside optimizations for payload indexing and resource utilization.
v1.13.43 fixes1 featureThis release introduces a strict mode for setting the maximum number of points in a collection and includes several bug fixes related to HNSW indexing, replica management, and consensus deadlocks.
v1.13.35 fixes8 featuresThis release focuses on stability and performance improvements, notably enabling consensus compaction by default and enhancing data consistency by cleaning up old point versions during updates. Several bugs related to panics and data flushing have also been resolved.
v1.13.22 fixes1 featureThis release introduces GPU support fallback for devices lacking half-float capabilities and resolves critical bugs related to blob storage panics and optimizer point reuse.
v1.13.18 fixesThis release focuses on stability and performance, including improvements to segment merging and numerous bug fixes addressing potential panics across HNSW, payload storage, and memory mapping.
v1.13.02 fixes4 featuresThis release introduces significant performance improvements, including GPU support for HNSW indexing and switching to mmap storage for payloads and sparse vectors. New features include runtime resharding and a strict mode for collection operations.
Common Errors
StatusRuntimeException2 reportsThe `StatusRuntimeException` in Qdrant often arises from network instability or resource limitations causing gRPC connections to be prematurely closed during data transfer. Mitigate this by increasing gRPC keepalive settings on both client and server, and ensure sufficient resources (CPU, memory, network bandwidth) are available to Qdrant. Additionally, check network infrastructure (firewalls, proxies) for any settings that might prematurely close long-lived connections since firewalls have defaults to close connections.
TranscendBizException1 reportThe "TranscendBizException" often wraps gRPC errors like "INTERNAL: Encountered end-of-stream mid-frame", indicating network instability or resource exhaustion during communication with the Qdrant server. Address this by increasing Qdrant server resources (CPU, memory) to handle the load, and ensure a stable network connection between your application and the Qdrant service, potentially by using retry mechanisms with exponential backoff in your client code. Consider adjusting gRPC keepalive settings to prevent idle connections from being dropped.
ResponseHandlingException1 reportResponseHandlingException in Qdrant usually indicates a mismatch between the expected and actual format of the server's response, often due to schema changes after a client update or an incorrect Qdrant Cloud version being used. To fix it, ensure your qdrant-client version is compatible with your Qdrant server (either self-hosted or Qdrant Cloud) and try updating or downgrading the client accordingly; verify the client's data serialization matches the server's expected format if the versions appear to be compatible.
IllegalArgumentException1 reportThe "IllegalArgumentException" in gRPC within qdrant, manifesting as "INTERNAL: Encountered end-of-stream mid-frame", usually indicates a mismatch of data sizes or formats between the client and server. A common fix involves ensuring the client sends properly sized and formatted data according to the defined protobuf schemas used by Qdrant. Specifically, verify that vector embeddings are of the correct dimensionality and data type, as discrepancies can lead to incomplete transmission errors.
PointIdError1 reportPointIdError in Qdrant usually arises from attempting operations on point IDs that don't exist within the collection or trying to insert a point with an ID that already exists. Fix this by carefully verifying point ID existence before deletion/update operations and ensuring unique ID generation for new points during insertion. If using UUIDs, confirm proper UUID generation to eliminate ID collisions.
ToolExecError1 reportToolExecError in Qdrant often arises during Rust builds due to missing system dependencies required by its crates or an outdated Rust toolchain. Ensure you have the necessary system libraries (like build-essential on Debian/Ubuntu, or their equivalents on Fedora/other distros) installed and that your Rust toolchain is up-to-date (using `rustup update`). Specifically for Fedora, install `cmake` and potentially other development tools using `dnf groupinstall "Development Tools"`.
Related AI & LLMs Packages
AutoGPT is the vision of accessible AI for everyone, to use and to build on. Our mission is to provide the tools, so that you can focus on what matters.
Get up and running with OpenAI gpt-oss, DeepSeek-R1, Gemma 3 and other models.
🦜🔗 The platform for reliable agents.
The most powerful and modular diffusion model GUI, api and backend with a graph/nodes interface.
LLM inference in C/C++
GPT4All: Run Local LLMs on Any Device. Open-source and available for commercial use.
Subscribe to Updates
Get notified when new versions are released