L

ListBench v2.5.13

Vision-first listing core & diagnostics (contract harness)

ModeVision → OCR → Core
Contract2.5.13
Capsv2.5.0
Backend: media.php (contract)

Upload artwork

Drop in final art files; ListBench handles vision, OCR & listing core.

Step 1 · Ingest
⤒
Drop PNG/JPG art here

Vision-only first pass, then conditional OCR if text is detected.

Uploaded images No images yet
Vision model: gpt-4o · OCR: Tesseract

Listing-core diagnostics

Per-image base title, description, tags & evaluation.

Step 2 · Review
Ready Drop an image to run vision & listing core.
Per-image listing cores 0 results
No listing cores yet. Upload artwork on the left to see vision output, OCR text, and a SANZ-product, SANZ-platform base listing for each image. 
Ready.
• Contracts & caps loaded.
• Vision (gpt-4o) required.
• OCR (Tesseract) runs only when vision reports TEXT-PRESENT: yes.
• Listing core enforces title/description/tag bands and slogan-in-title rules.
Note: This harness focuses on artwork-only listing cores. The per-image product, platform, and holiday/event selectors will feed adaptation downstream.