Find real Hugging Face model repositories
Ask for models for a task or by name and Zeplik searches the live model hub, then answers with real repositories, their exact repo ids, and live download and like counts. This matters because a language model on its own will invent a repo id or misstate a model's popularity from a stale snapshot, sending you to a page that does not exist. Here every result is grounded in the live hub, and when nothing matches Zeplik says so rather than fabricating a model.
The lookup is free; you pay only the normal cost of the answer, shown before you send.
How it works
Describe the model or task
Ask in plain language, like 'popular open models for speech-to-text' or 'find the Llama 3 8B instruct repo'. A model name or a task is enough for Zeplik to run the search.
Zeplik searches the live hub
It queries the live model hub, not the model's memory, and treats each result as source data handled as untrusted content, so nothing in a model card can hijack the answer.
Read the grounded result and open it
The answer returns real repositories with their exact repo id and live download and like counts, so you can open the page, confirm the details, and compare options in the same conversation.
Worked examples
Real prompts you can open in the composer with the text ready to review. Nothing sends on its own. Swap in your own details.
Prompt: What are the most downloaded open models for text embeddings?
Zeplik searches the live hub and returns real repo ids with current download counts, so a shortlist reflects actual popularity, not a stale guess.
Open this in ZeplikPrompt: Find the official Llama 3.1 8B Instruct repository and its download count
It resolves the exact repo id and reports the live count, so you land on the real page instead of an invented path.
Open this in ZeplikPrompt: Someone recommended a model called 'mistral-super-7b-v9'. Can you verify it exists?
Verification is the point: it checks whether the repo resolves on the hub and tells you plainly if it cannot be confirmed.
Open this in ZeplikWhat it can and cannot do
The lookup is precise about what it can verify and honest about its edges. Here is the plain posture.
Search the live hub by model name or task and get real repositories with their exact ids.
Each result comes back with current popularity signals read live, so a comparison reflects the present, not a stale snapshot.
Paste a repo id and Zeplik checks whether it resolves on the hub, catching a fabricated or mistyped reference.
If no model matches, Zeplik says so plainly instead of inventing a repository that does not exist.
It reports what the hub records, like counts and metadata, not third-party benchmark rankings. For those, check the model card or a benchmark leaderboard.
It surfaces the stated license where recorded, but it does not advise whether a license fits your commercial use. Review the license terms and consult counsel if needed.
Grounded, and honest by default
- It cites real repositories. Every result comes back with its exact repo id and live counts, so you can open the real page instead of trusting the model's memory.
- It will not invent a model. When nothing matches or a repo id cannot be confirmed, Zeplik says so rather than fabricating a plausible-looking repository.
- Model-card text cannot hijack the answer. Each hub record is treated as untrusted content: Zeplik reads and reports it but never obeys instructions embedded inside it.
Frequently asked questions
- Why not just ask a chatbot to recommend a model?
- Because a language model on its own will confidently give a repo id or a popularity figure from a stale snapshot, and can invent a repository that does not exist. This tool searches the live hub and returns real repositories with current download and like counts, and it refuses to fabricate one when nothing matches.
- Which hub does it search?
- It queries the live public model hub. Give a model name or a task and Zeplik returns matching repositories with their exact ids and live popularity signals.
- Can it verify a repo id I was given?
- Yes. Paste a repo id and Zeplik checks whether it resolves on the hub and summarizes the repository, so you can catch a fabricated or mistyped reference before you rely on it.
- Does it tell me which model is best?
- It surfaces real matches with live popularity signals so you can shortlist, but it does not provide independent benchmark scores or judge fitness for your use. Read the model cards and, for licensing questions, review the terms. Zeplik is clear about these limits.
- Does the search cost credits?
- The model lookup itself is free. You pay only the normal model-token cost of the answer written around it, shown before you send, like any other message.
Related skills
More on Zeplik
Find models that actually exist
Search real model repositories with their exact ids and live download counts, grounded in the live hub. Free to start.