For institutional investors
Teach a model to filter your news.
Monitor more information with an open-weight model trained on your judgment.
Methodology outperforms frontier modelsFor institutional investors
Monitor more information with an open-weight model trained on your judgment.
Methodology outperforms frontier modelsStep 1 · Your watchlist
Step 2 · Training labels
Train your model to decide whether stories like this belong in your newsfeed.
Step 3 · Ready & save
There is enough signal to train a first version of your news filter.
Step 4 · Live run
Connecting to the run…
The chart begins when Tinker returns the first completed mini-batch.
Nothing here is estimated except ETA. Loss, throughput, timing, tokens, checkpoint checks, and story classifications update only after the server receives a completed Tinker result.
The run hit an error. Your labels and coverage are saved.
Model
Your Keep and Skip decisions become the next version of your filter.
Your decisions will appear here.
What you follow
Choose the tickers that should appear in your feed.
Account