Should you pay for this lead?
Upwork, Thumbtack, Bark and PRO360 charge you before you even talk to the client. Paste the job post. Gemma 4 reads it on your own GPU, plain code checks every quote it gives back, and the math decides.
The lead
Your numbers
In USD. Change them and the verdict updates at once, without running the model again.
How it decides
- Gemma 4 E2B reads. It copies the budget, the pieces of work and any warning signs out of the post, each with an exact quote. It is not asked for an opinion.
- Code checks. A quote that is not in the post is thrown out, along with the claim it was supposed to support. Amounts are read from the quote by code, not taken from the model.
- Code decides. Hard no's (on site, account farming, paid only on success, unpaid tests) end it. Otherwise the cost of bidding (lead fee plus your unpaid proposal time) is divided by what the job pays above your floor. That is the share of these leads you must win to break even. If your real win rate is higher, pay.
Nothing you type leaves your browser. The model (about 2.8 GB) downloads once from Hugging Face and WebLLM's runtime file from GitHub; after that it works offline. Recorded runs are real outputs from the same model, saved so you can see the tool without the download. Examples are made up or rewritten from my own notes; each one says which.