Skip to main content

← Dracon Bench leaderboard

nvidia/nemotron-3.5-lightning:free

Tested First tested 2026-11-11 · OpenRouter free lane

Nvidia Nemotron 3.5 Lightning free lane: IFEval-50 89.8%, GSM8K-50 70.0% with zero failures, SVG 55/250 over 5 judged tests. Best IFEval on the board so far, weakest GSM8K. The clock brief scored 0: the model emitted JavaScript-generated SVG instead of static shapes, and scripts are stripped before render.

Scores

SuiteScore
IFEval-50 prompt-strict89.8% (n=49)
IFEval-50 instruction-strict93.7% (n=49)
GSM8K-5070.0% (35/50)
SVG blind rubric55/250

Artifacts

Blind-judged by nvidia/nemotron-3.5-super-vl:free: the judge saw the brief and the rendered image only, never the model name. Briefs are frozen; expand one to read it.

beach · 21/50

Brief fidelity (3/10)
Most headline elements are missing or unrecognizable: no dolphins, birds, clear sailboat, shimmering reflection path, or leaning palm silhouette, with only the sun and sandy foreground present.
Construction (6/10)
Shapes are simple geometric primitives (circular sun, rectangular trunk, semi-circle clouds) that are cleanly assembled without melting, duplication, or broken anatomy.
Composition (4/10)
The layout places the sun right and palm left with a flat horizon, but the frame feels empty and lacks convincing depth, relative scaling, or atmospheric perspective.
Detail finish (3/10)
The image is extremely minimal with no shading beyond the sky gradient, missing textures and small touches, and the palm trunk is cut off by the left edge.
Visual appeal (5/10)
The warm orange-to-teal palette is pleasant, but the sparse, abstract execution lacks charm and fails to evoke the requested sunset beach scene.
Brief

Draw a beautiful sunset beach scene: a large sun sitting low over the sea with a shimmering reflection path on the water, a leaning palm tree silhouette on the left, two dolphins arcing out of the waves, a small sailboat on the right, birds in the sky, sandy foreground with a starfish. Warm oranges and pinks fading to teal. Coherent perspective: distant objects smaller and hazier than near ones. Output: exactly one self-contained SVG (viewBox "0 0 800 600", no external references) inside a single ```svg fenced block. Short prose outside the fence is allowed.

clock · 0/50

Brief fidelity (0/10)
The rendered image is completely blank; none of the requested clock face, numerals, hands, or DRACON text are present.
Construction (0/10)
There are no assembled shapes, hands, or object parts visible in the empty frame.
Composition (0/10)
The composition is an empty white canvas with no layout, framing, or centered elements.
Detail finish (0/10)
No detail, shading, or finish is present; the image contains no visible SVG content at all.
Visual appeal (0/10)
A blank white image has no coherent palette, linework, or charm and would not be considered good.
Brief

Draw a round wall clock face showing exactly 10:10: a circular dial with all 12 hour numerals in the correct positions, minute tick marks, an hour hand pointing just past 10, a minute hand pointing exactly at 2, and a thin second hand pointing at 6. The word DRACON in small capitals under the center. Dark rim, light face, flat vector style, centered composition. Output: exactly one self-contained SVG (viewBox "0 0 800 600", no external references) inside a single ```svg fenced block. Short prose outside the fence is allowed.

duo · 21/50

Brief fidelity (4/10)
Headline elements are missing, no visible paws on handlebars, no raccoon holding on, no full bodies, and no clear leaning turn, so this is capped at 4 despite the sun, clouds, shoreline, and spray being present.
Construction (3/10)
The figures are reduced to floating heads with no bodies, paws, or coherent jet-ski anatomy, making them look assembled only in the most rudimentary sense.
Composition (5/10)
The characters sit low in the frame with a simple horizon and open sky, creating a readable but empty layout with minimal perspective.
Detail finish (4/10)
Flat colors with small touches like whiskers and a raccoon mask, but no shading, textures, or refined finishes, and the jet ski is just a curved red shape.
Visual appeal (5/10)
The palette is bright and coherent with confident simple linework, but the scene lacks the comical action and charm promised by the brief.
Brief

Draw a comical scene: a dog driving a jet ski at speed, paws on the handlebars, with a raccoon riding behind holding on, both leaning into a turn. Sea spray behind the jet ski, sun and clouds in the sky, a distant shoreline. Both characters must be clearly doing something, not just posed side by side. Bright cartoon vector style. Output: exactly one self-contained SVG (viewBox "0 0 800 600", no external references) inside a single ```svg fenced block. Short prose outside the fence is allowed.

monalisa · 6/50

Brief fidelity (0/10)
None of the requested Mona Lisa elements are present: no seated woman, three-quarter view, dark dress, folded hands, enigmatic smile, parted hair, or background landscape with river and bridge.
Construction (1/10)
There are no bodies, faces, or hands; the image consists of abstract geometric blobs that do not assemble into recognizable anatomy or objects.
Composition (2/10)
The layout places random central shapes above green hills with no portrait composition, relative sizing, perspective, or use of the frame for a bust.
Detail finish (1/10)
No shading, textures, neckline details, or small touches from the brief appear; the artwork is effectively unfinished relative to the request.
Visual appeal (2/10)
The muted palette is coherent but the image lacks charm, confidence, and recognizability as a Mona Lisa portrait.
Brief

Redraw the Mona Lisa as a flat vector portrait, bust composition: a seated woman in three-quarter view facing the viewer, dark dress with neckline detail, hands folded at the bottom edge, soft enigmatic smile, hair parted in the middle. Behind her a hazy landscape with a winding river, a bridge on the right, and blue mountains. Muted greens, browns, and smoky blues. Output: exactly one self-contained SVG (viewBox "0 0 800 600", no external references) inside a single ```svg fenced block. Short prose outside the fence is allowed.

pelican · 7/50

Brief fidelity (1/10)
The pelican is not recognizable; the beak with pouch, long neck, folded wing, and webbed feet are all missing, so headline elements are absent and this criterion is capped low.
Construction (1/10)
No assembled anatomy is present, only disconnected geometric blobs (a circle, a triangle, a cut-off oval) with no body, head, or limb structure.
Composition (2/10)
The sky/sea/ground bands occupy the frame, but the subject is cut off at the right edge and lacks placement or scale, making the composition incomplete.
Detail finish (1/10)
There is no shading, texture, or finish; the large brown shape is cut off by the frame edge and stray shapes (floating circle, triangle) appear unresolved.
Visual appeal (2/10)
The flat palette is coherent but the image reads as abstract fragments rather than a charming or confident pelican illustration.
Brief

Draw a brown pelican standing in side profile, facing left, full body visible: large beak with throat pouch, long neck, folded wing, webbed feet on flat ground. Simple sky background with a hint of sea on the horizon. Flat vector style, clean shapes, limited palette. Output: exactly one self-contained SVG (viewBox "0 0 800 600", no external references) inside a single ```svg fenced block. Short prose outside the fence is allowed.

Compare with

Evidence and notes

  • Clock 0/50 is genuine: SVG draws via <script>, blank once sanitised (verified by independent render)
  • 7 of 8 IFEval failures recovered at 8k; 1 prompt returned empty content, counted as failure
  • Clock artifact recovered at 32k after a first-attempt reasoning burn