Bench leaderboard
This table ranks models on markup, not on whether the output works.
No run has been executed and graded yet. see verified examples →
No run has been executed and graded yet. see verified examples →
Auto-scored matrix across 13 runs · 1 model · 6 skill kinds. Scoring rubric: validity + fidelity + structure + depth + cleanliness (max 10/step).
Score over time
Dots = individual steps. Solid lines = rolling-3 average per model. As we swap models + skills the line tells the story of where we got better — or worse.
qwen-image (29)
Axis breakdown over time
Each rubric axis (max 2) plotted independently. Useful for diagnosing — e.g. fidelity climbs when subject-token recall improves, structure climbs when models start emitting proper semantic tags.
validity fidelity structure depth cleanliness
Model × skill (avg of 13 runs)
| Model | code-app | image-gen | project-dna | slide-deck | style-guide | website | overall |
|---|---|---|---|---|---|---|---|
| qwen-image | 0.0 /10 (1) | 8.0 /10 (12) | 7.5 /10 (4) | 9.0 /10 (1) | 10.0 /10 (1) | 9.3 /10 (10) | 8.2 /10 |
All graded runs
r-1790793915767-gfjvx 16 / 20 qwen-image r-1790182166534-ksei1 17 / 20 qwen-image r-1789212998836-3v450 17 / 20 qwen-image r-1789176550407-zvah5 18 / 20 qwen-image r-1789166894383-2sz6g 20 / 20 qwen-image r-1789124656615-6iofk 8 / 20 qwen-image r-1789123948434-cbqg4 17 / 20 qwen-image r-1789123565073-lz356 18 / 20 qwen-image r-1789115880572-82i5l 18 / 20 qwen-image r-1789114658053-o2lbb 28 / 30 qwen-image r-1789088913677-u7qeo 28 / 30 qwen-image r-1789088863428-yzaq7 9 / 10 qwen-image r-1788504869178-mz6dl 24 / 40 qwen-image