recursive/flint-v1.0:free

Tested First tested 2026-11-09 · OpenRouter free lane

Recursive's Flint v1.0 is the spike's calibration anchor: its 93.3% IFEval-50 lands within 2 points of the vendor-published full IFEval figure (95.1%), which is what a correct harness should do on a 50-prompt sample. GSM8K-50 88.1%, SVG total 144/250 with the clock (37/50) as its best test.

Scores

Suite	Score
IFEval-50 prompt-strict	93.3% (n=45)
IFEval-50 instruction-strict	92.9% (n=45)
GSM8K-50	88.1% (37/42)
SVG blind rubric	144/250

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 · 31/50

Brief fidelity (4/10): Missing the two dolphins entirely and the starfish is only a tiny abstract mark, so headline elements are absent, capping fidelity at 4.
Construction (7/10): All visible objects are simple geometric or silhouette shapes that are cleanly assembled with no anatomical errors or melted forms.
Composition (8/10): The palm anchors the left, the sun centers the horizon, the boat balances the right, and atmospheric perspective is respected with distant haze and near foreground.
Detail finish (5/10): The scene is flat and minimal with a reflection path and birds, but lacks textures, shading, and a recognizable starfish.
Visual appeal (7/10): The warm-to-teal gradient and confident silhouettes create a pleasant, if simple, postcard-like charm.

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 · 37/50

Brief fidelity (7/10): The clock displays 10:10 with all 12 numerals correctly ordered and DRACON centered below, but the hour hand points directly at 10 rather than just past it, and the minute tick marks are rendered as an incomplete dotted pattern rather than clear marks around the dial.
Construction (9/10): All clock components, the circular rim, face, numerals, and three hands, are cleanly assembled with no melted, duplicated, or malformed parts.
Composition (9/10): The clock is perfectly centered within the 800x600 frame with balanced proportions between the dark rim, light face, and numeral placement.
Detail finish (5/10): While the flat style is clean and the red second hand is a thoughtful accent, a stray dotted line extends beyond the clock face at bottom-right and the minute tick marks appear as an incomplete dotted arc rather than distinct marks around the full dial.
Visual appeal (7/10): The dark rim, cream face, black numerals, and red accent create a coherent, confident palette, though the stray dotted line slightly undermines the polished look.

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 · 26/50

Brief fidelity (3/10): The headline elements of a jet ski with handlebars are missing, the image shows a kayak with a paddle, so the score is capped and set at 3.
Construction (6/10): All bodies and object parts are clearly assembled vector shapes without melted or duplicated anatomy, though rendered in a simple style.
Composition (6/10): The horizon divides the frame, the boat is the clear focal point left of center, and the sun and clouds balance the upper space.
Detail finish (5/10): Flat colors, black outlines, sea spray, sun rays, and clouds are present, but there is little shading, texture, or fine finishing.
Visual appeal (6/10): The coherent blue-and-orange palette and confident outlines create simple charm, though the image is minimal and not especially polished.

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 · 26/50

Brief fidelity (4/10): The figure has a dark dress, middle-parted hair, smile, mountains, and river, but it is frontal rather than three-quarter, the hands are not clearly folded at the bottom edge, and the bridge is only an ambiguous brown rectangle, so headline elements are missing.
Construction (6/10): The portrait is built from clean geometric shapes rather than melted forms, but the anatomy is highly stylized with an oversized head, no neck, and two floating circles for hands, making it only loosely assembled.
Composition (6/10): The subject is centered against a layered background of mountains and river, using the frame effectively, though the disproportionately large head and flat perspective create an awkward visual hierarchy.
Detail finish (5/10): The artwork uses flat muted colors with no shading or texture, includes a simple smile and hair part, and shows no broken markup or cut-off edges, but the two orange circles at the bottom are ambiguous and under-detailed.
Visual appeal (5/10): The palette of muted greens, browns, and smoky blues is coherent and the linework is confident, but the naive abstraction lacks the charm or recognizability of a successful Mona Lisa reinterpretation.

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 · 24/50

Brief fidelity (4/10): The bird is brown, in left-facing side profile with a long neck and full body visible, but the throat pouch is only a detached dot and the webbed feet are missing/not grounded, so headline elements are absent and this is capped at 4.
Construction (5/10): The body, neck, head, beak, wing shape, and tail are assembled from clean shapes, but the feet are floating unconnected ovals and the pouch appears as a stray circle rather than attached anatomy.
Composition (6/10): The pelican occupies the center of the frame with a clear horizon dividing sky, sea, and ground, using the 800x600 space effectively without crowding.
Detail finish (4/10): The style is flat and minimal with a limited palette, but there is no shading, texture, or finish beyond basic shapes, and the detached pouch/feet look unfinished.
Visual appeal (5/10): The coherent brown-blue-tan palette is pleasant, but the illustration is crude and lacks charm or confident detail, making it only modestly appealing.

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

Vendor-published full IFEval figure 95.1% (Recursive model announcement, Nov 2026)
8 GSM8K and 5 IFEval prompts unrecoverable: reasoning trace exhausted the token budget, counted as failures