AI Devs Shift to Qdrant, Weaviate for Performance DBs

Package download data from AI-Buzz covering the past 30 days shows Chroma losing ground in both major developer ecosystems while Weaviate and Qdrant continue to grow. The numbers are unambiguous: Chroma's npm downloads dropped 15.4% and its PyPI downloads fell 2.8%, while both competitors posted gains across the board. For a tool that built its reputation on being the easiest vector database to get started with, the trajectory is worth examining closely.

Weaviate's Scale, Qdrant's Momentum

Weaviate recorded 23,655,806 PyPI downloads over the period, a 2.5% increase on an already large base, according to AI-Buzz tracking data. Its JavaScript library grew 6.0% to 1,833,481 downloads. At this volume, even modest percentage growth represents a substantial number of new installations.

Qdrant is the more interesting story. Its PyPI downloads surged 6.4% to nearly 11.8 million (AI-Buzz data), while npm downloads rose 4.2% to 1.4 million. Growth across both ecosystems at roughly the same rate suggests a broad, consistent expansion rather than a spike in one area. Qdrant also leads on GitHub engagement: its repository has over 29,000 stars, compared to Chroma's 26,407 and Weaviate's 15,706.

A community that large tends to generate its own momentum through tutorials, integrations, and word-of-mouth.

Chroma's 15.4% npm Drop

The JavaScript decline is the sharpest signal in the data. Chroma's npm downloads fell 15.4% to 716,069, based on AI-Buzz's developer adoption dashboard. PyPI downloads declined more modestly, down 2.8% to 11,852,788, but the direction is the same in both ecosystems.

The JavaScript drop is particularly telling because full-stack developers building user-facing applications tend to be early movers when infrastructure isn't meeting their needs. A review of Hacker News discussions surfaces recurring concerns about Chroma's scaling behavior and production readiness. Qdrant is written in Rust; Chroma is not. That difference gets noticed when teams start stress-testing their stacks.

The sentiment data adds context. Despite generating 31 Hacker News mentions, the highest of the three, Chroma's positive sentiment score sits at just 21.0%, according to AI-Buzz. Qdrant sits at 42.0% positive across 30 mentions. Weaviate leads at 50.0%.

High mention volume with low positive sentiment is a specific pattern: people are talking about Chroma, but often to flag problems.

Why Production Pressure Favors Qdrant and Weaviate

Chroma's early appeal was straightforward: get a vector database running in a few lines of Python, no infrastructure overhead. That's a real advantage for prototyping RAG pipelines or semantic search demos. The problem is that the same properties that make a tool easy to start with, minimal configuration, in-process operation, limited tuning options, can become friction when an application needs to scale.

Qdrant markets itself explicitly as a high-performance vector similarity search engine, a positioning detailed in its official documentation. Its $28M Series A in early 2024 gave it resources to accelerate enterprise features. Weaviate's sustained relevance comes partly from its breadth: GraphQL and REST APIs, a modular architecture, and a feature set that accommodates complex production requirements. Neither tool is as frictionless as Chroma for a first project.

That trade-off appears to be one more teams are willing to make.

What the Numbers Don't Settle

Download counts measure adoption, not satisfaction or retention. A project that installs Qdrant once and never uses it still registers as a download. Chroma's absolute PyPI numbers, nearly 11.9 million, remain large, and the Python decline of 2.8% is not a collapse. It's possible that Chroma's core user base, developers running local or small-scale deployments, remains stable while the projects pushing toward production are the ones leaving.

What the data does show clearly is that Chroma is not growing while its two main competitors are. Over a 30-day window that's a trend, not a verdict. But if the npm decline accelerates or the PyPI gap widens, the question for Chroma's team becomes whether the product's positioning needs to change, or whether the execution does. The answer matters for anyone currently choosing infrastructure for an AI application they expect to still be running in two years.