Quelllm MCP
MCP server to list, compare and size open-weights LLMs from the quelllm.fr catalog
- Pricing
- Open source
- Platforms
- CLI, API, self-hosted
- Version
- 1.1.0
- Listing updated
- today
Similar MCP servers
About Quelllm MCP
Quelllm MCP is an open-source MCP server for ai models and llm tooling. It is available on CLI, API and self-hosted.
The README describes this as a bridge between AI clients and the quelllm.fr catalog of open-weights language models, which it says covers more than 190 models. Once connected, a client such as Claude Code or Cursor can filter models by origin, family or parameter count, pull a full record for a single model, and run side-by-side comparisons.
Two estimation tools are the practical focus: one computes the VRAM needed at a chosen quantization and suggests GPU or Mac tiers, and another compares monthly API spend at a given token volume against self-hosted hardware. The README notes that the cost figures rely on hardcoded assumptions last checked in May 2026 and should be verified before any purchase.
The server is aimed at developers and hobbyists choosing a local model for their hardware, and it pulls its data from the quelllm.fr public API under a CC BY 4.0 licence. The project is MIT-licensed and installed from source with pip or uvx.
Tools this server provides (6)
Taken from the project documentation; the running server may differ.
- list_models
- List models with filters for origin code, family and maximum parameters in billions.
- get_model
- Return the full record for one model.
- compare
- Side-by-side comparison of two models with a verdict.
- estimate_vram
- VRAM in GB at a chosen quantization with recommended GPU or Mac tiers.
- estimate_cost
- Estimated cost in EUR across API providers and self-hosted hardware.
- search_models
- Fuzzy search by name, family, tag or author.
What it does
- Model catalog filtering
- List models by origin code, family or maximum parameter count.
- Side-by-side comparison
- Compare two models with a verdict, using their catalog IDs.
- VRAM estimation
- Estimate memory needed at a chosen quantization with suggested GPU or Mac tiers.
- API versus self-hosted cost
- Estimate monthly EUR cost across API providers and local hardware from token volumes.
- Fuzzy model search
- Search the catalog by name, family, tag or author.
- Cached local data
- Catalog responses are cached locally for one hour to limit repeated requests.
Good for
- Checking which open-weights models fit a specific GPU or Mac before downloading them
- Comparing two candidate models from inside an MCP-enabled coding assistant
- Estimating whether self-hosting is cheaper than paying a cloud API for a given monthly token volume
- Finding a model in a given family or size range without leaving the chat client
Pricing
Open source
Free MIT-licensed server; the catalog data is CC BY 4.0.
Prices change. Check quelllm.fr before you buy.
Quelllm MCP FAQ
How do I install the server?
The README says it is not yet on PyPI and gives install commands using pip from the GitHub repository, or zero-install with uvx.
Which clients does it support?
The README gives configuration examples for Claude Code, Claude Desktop, and says Cursor, Continue and Cline accept the same JSON config.
Are the cost estimates current?
No. The README says the estimate_cost tool uses hardcoded assumptions last checked in 2026-05 and may be outdated, so provider pricing should be verified.
What license is it under?
The README states the project is released under the MIT licence.
Is Quelllm MCP free?
Yes. Quelllm MCP is open source, so you can use and self-host it for free.
What are the best alternatives to Quelllm MCP?
The closest alternatives to Quelllm MCP are Local AI MCP, Model Tested, Open Weights Index, AI Model Analyzer and MCP.
Where this comes from
- Description: written from the project's README, the project's website and its MCP Registry entry.