GitHub radar
Pick an open model for scoring text, ranked by un-gameable tests
This is a catalog of 100+ open models trying to reproduce TypeSafe's paid Jev model for scoring or labeling text. Instead of digging through repos yourself, you pick a ready model from the ranking.
The site compares over a hundred open models that output a score or label for text (no written reply), and ranks them by benchmarks.
When it helps
Useful if you need cheap, fast text classification — like sorting tickets or comments by one trait. Won't help if you need proof a model matches the paid Jev — the site can't verify that.
Pros
- Ranks models on private tests that can't be memorized in advance
- Saves time — skip digging through a hundred+ repos yourself
- Shows both public and private scores side by side
- Runs right in the browser, nothing to install
Cons
- Doesn't prove a model matches TypeSafe's paid Jev
- The site builds no models itself — just a catalog and ranking
- Author still recommends testing the top model before production use
How to set it up — step by step
- 1Open https://huggingface.co/spaces/multimodalart/jev-decision-index in your browser.
- 2Send your agent the text from the block below to find a model for your task.
- 3Open the Hugging Face link to the model the agent gives you.
- 4Compare its public vs private score on the model's own report page (?model=...).
- 5Test the model on your own text samples before using it in production.
Text for your agent
Copy this and send it to your agent — Claude Code, Codex, any of them:
Open this page https://huggingface.co/spaces/multimodalart/jev-decision-index and tell me which open model currently ranks best on the private tests for sorting text, like tickets or comments, by a single trait — and give me its link on Hugging Face.
Nothing to install - it runs in the browser.
Open on Hugging Face▌ More finds