QMedia

QMedia is an open-source search and question-answering system for creators managing text, images, and short videos. It suits teams willing to deploy services and models themselves, especially when private, mixed-media collections matter. Its breadth is valuable, but operating the separate RAG dependency and enabling video processing require deliberate setup.

Overview

QMedia combines multimedia retrieval with question answering rather than treating each format as a separate library. It searches and extracts information from text, images, and short videos through a multimodal RAG system, allowing a question to draw on more than one kind of source.

The web interface presents results as cards containing source references and separate breakdowns of image, text, and short-video information. That presentation is useful for creators who need to inspect the material behind a result, not merely receive a generated response. QMedia can also be deployed locally for offline search and Q&A over private data. Local operation provides control over processing location, but it is not a stated security certification or compliance status.

The project is released under the MIT License. Its README directs users to issue reporting and requests and points to Discord and WeChat communities for discussion.

Key features

• Multimedia search and extraction: Text, images, and short videos can be searched in one system. This is a stronger fit for creator archives with mixed media than a product limited to video or image indexing.
• Multimodal Q&A: The RAG layer combines information across supported formats for questions and answers. The RAG service requires the separate mm_server service, so deployment is not a single-component installation.
• Inspectable content cards: Cards show sources and break down image, text, and short-video information. That gives users a way to follow the evidence and understand which media contributed to a result.
• Separable services: The web service, RAG search and Q&A service, and image/text/video model service can be deployed independently. This supports architectures that separate user interface, retrieval, and model workloads, while adding operational coordination.
• Model choices: Documentation covers local Ollama model switching and names CLIP, BGE, Qanything OCR, LLaVA, and Faster Whisper among the options. The range spans visual embeddings, OCR, vision-language work, and speech transcription, giving self-hosters room to match models to their collection.
• Model-service API: The standalone multimodal model service exposes image and text encoding, video transcription, and image OCR through an API. It can therefore serve other applications as well as QMedia's own search workflow.
• Configurable external search: The RAG service supports Google content search, which is configurable and disabled by default. Teams can leave that integration off when they want a purely local workflow.
• Video control: Video-service use is configurable and off by default in the RAG server configuration. That avoids enabling video processing until it is needed, but video-focused deployments must change the setting.

Pricing

QMedia — free: The software is free and open source under the MIT License. There are no paid tiers, seat charges, quotas, trial period, or renewal terms stated. This is the natural choice for creators and technical teams that can provide their own deployment and model-service resources. The trade-off is that the free offering is the project itself: it does not describe a hosted plan that removes service administration.

Platforms

QMedia is offered through a web interface, an API, and self-hosted deployment. The self-hosted route supports local, offline work over private data; the API route is aimed at integrating the multimodal model capabilities into another system. Because the services can be deployed separately, platform planning includes the web, RAG, and model-service components rather than just a client install.

Who it's for

QMedia is for content creators, archives, and development teams with text, image, and short-video material that they want to search together and question conversationally. It is particularly appropriate when keeping private data on local infrastructure is important, or when an API for encoding, OCR, and transcription is useful. It is less suitable for users seeking a managed, ready-to-run hosted search product, or for teams unwilling to operate the required mm_server dependency and configure video services.

Pros and cons

Pros

• One workflow for three media types: Text, images, and short videos are searchable and available to the same Q&A system.
• Evidence-oriented results: Source-linked cards and media breakdowns make answers easier to examine.
• Local privacy control: Offline deployment supports private-data workflows without requiring a hosted service.
• Flexible integration: Separate services, Ollama switching, and a model API accommodate custom deployments.
• No software fee: MIT-licensed open-source software avoids subscription pricing.

Cons

• More deployment work: The RAG service depends on mm_server, and the architecture separates three service areas.
• Video is not automatic: Video-service use is off by default and must be configured for video processing.
• Self-hosting responsibility: The project does not provide a stated security certification or compliance status, so those controls remain the deployer's responsibility.

Alternatives

Choose FrameQuery instead if you want freemium access across API, macOS, web, and Windows, with a free tier that includes metadata search but no monthly indexing hours; indexed video is required for its face, voice, and AI-tag searches.

VGG Image Annotator is the simpler free, cross-platform option when annotation is the priority rather than multimodal RAG and Q&A; it has no paid tiers or stated usage limits.

Pick DISKO when a macOS/web production workflow with screenwriting, shot lists, review links, and a 2 GB free cloud tier is more relevant than self-hosted multimedia search.

Cinestar is another free, Linux/macOS choice with an MIT open-source core, while its local AI services require Ollama.

For a hosted API with explicit video quotas, Twelve Labs offers a free plan with 600 minutes of video, 90-day index access, and defined per-index limits.

Verdict

Choose QMedia if you need open-source, local-capable search and Q&A across text, images, and short videos, with model APIs and deployment control. Look elsewhere if you want a managed experience or cannot accommodate the separate RAG dependency and video configuration.