[Submitted on 21 Jul 2026]

View PDF HTML (experimental)

Abstract:Instruction-based vector editing requires two capabilities: making a requested change and leaving everything else alone. The second is easy to miss when an output is judged only as a raster image. We introduce Vector-Bench, a compact, difficult benchmark of 40 SVG repair tasks. Each task pairs a corrupted SVG program with an author-written visual instruction, a hidden target program, 5.05 annotated repairs on average, and an average of 60.55 protected objects. Instructions describe visible defects without exposing element identifiers, coordinates, color codes, or path data. We define a deterministic binary specification reward: requested repairs use attribute-aware perceptual tolerances, while unrequested rendering- or application-relevant structure must remain semantically unchanged and the result must be a valid SVG. Canonical target equality and stricter source fidelity are retained as diagnostics. Validity-gated repair progress, a near-complete tier, and valid-output Unintended Change Rate (UCR) explain partial outcomes. We evaluate 34 model endpoints (25 listed as open-weight, 5 inexpensive controls, and 4 frontier closed endpoints) over 1360 requests. The strongest endpoint reaches only 15.0% full specification success, despite 43.7% mean repair progress, showing that apparent repair progress and specification-faithful editing remain substantially different. All prompts, outputs, scoring code, costs, and per-task reports are released.

Submission history

From: Prannay Hebbar [view email]
[v1] Tue, 21 Jul 2026 12:44:39 UTC (358 KB)