Overview · R1–R180

Every round, every model, one page.

The complete index of blindtest.ai: 58 published rounds across eight disciplines. Rounds R31 and later were scored blind — reviewed without knowing the author; every answer is published under its model's name. Rounds R1–R30 are archive rounds with names attached. Scored rounds carry blind rubric scores (0–10 against a documented gold answer); creative and style rounds stay deliberately unscored. R1–R5 predate that rubric entirely — they carry the original 2026-07-17 editorial scores instead.

179rounds published — R1–R180, every raw answer browsable

13models compared · 8 local, 5 cloud

50blind-scored rounds (R31 and later), every answer named

All rounds newest first

Round	Task	Category	Format
Blind-first rounds · R31–R80
R180	Fix the broken references	editing-correction	blind
R179	Unify tense and perspective	editing-correction	blind
R178	Remove redundancy without losing facts	editing-correction	blind
R177	Make the numbers consistent	editing-correction	blind
R176	Fix exactly the errors, change nothing else	editing-correction	blind
R175	Noise flowfield painting additive particle trails	Code	blind
R174	Recursive fractal tree growing branch by branch	Code	blind
R173	Metaball lava lamp with heating and cooling blobs	Code	blind
R172	Raindrops sliding down a window pane	Code	blind
R171	Fireworks bursting into fading sparks	Code	blind
R170	Compress a how-to into exactly 5 numbered steps	summarization	blind
R169	Neutral summary of a biased product rant	summarization	blind
R168	Extract action items as a checklist	summarization	blind
R167	One-sentence TL;DR under 25 words	summarization	blind
R166	Faithful 3-sentence summary of a chaotic meeting transcript	summarization	blind
R165	Recursive backtracker carving a maze, then solving it	Code	blind
R164	Boids flocking with separation, alignment, cohesion and a predator	Code	blind
R163	Prim's algorithm growing a minimum spanning tree	Code	blind
R162	Breadth-first search flooding a random grid maze	Code	blind
R161	Selection sort vs shell sort racing side by side	Code	blind
R160	Critique a generic slogan in 2 sentences + give 2 better alternatives	Creative & Marketing	blind
R159	Micro-story: exactly 50 words (±5) with 3 mandatory elements	Creative & Marketing	blind
R158	5 email subject lines under 50 chars, no clickbait, no ALL-CAPS	Creative & Marketing	blind
R157	Rewrite the same product text for two different audiences, facts unchanged	Creative & Marketing	blind
R156	Invent 3 product names + taglines for a fictional object (no superlatives)	Creative & Marketing	blind
R155	Linked list with animated pointer relinking	Code	blind
R154	Ring buffer with producer and consumer at different rates	Code	blind
R153	Binary min-heap with bubble-up and sink-down	Code	blind
R152	Hash table with visible collision chains	Code	blind
R151	Binary search tree growing node by node	Code	blind
R150	Diff two near-identical config blocks	Structure & Data	blind
R149	Extract meeting invite into a fixed JSON schema	Structure & Data	blind
R148	Table QA: sum, filter, max over order data	Structure & Data	blind
R147	CSV with embedded comma to JSON array	Structure & Data	blind
R146	Repair broken JSON (4 precise errors)	Structure & Data	blind
R145	False-friend minefield German into English	Language & Translation	blind
R144	English apology into idiomatic Sie-German	Language & Translation	blind
R143	Bureaucratic German into plain English	Language & Translation	blind
R142	Register shift: casual complaint into formal	Language & Translation	blind
R141	Idiom-dense German into natural English	Language & Translation	blind
R140	Water sloshing in a tilting container	Code	blind
R139	Small n-body system orbiting a central star	Code	blind
R138	Two-source wave interference pattern	Code	blind
R137	Cloth mesh hanging and rippling in the wind	Code	blind
R136	Double pendulum tracing a chaotic path	Code	blind
R135	Conditional dice probability	Knowledge & Logic	blind
R134	Smallest number from three remainders	Knowledge & Logic	blind
R133	Three friends, three pets	Knowledge & Logic	blind
R132	Fermi estimate: a lifetime of drinking water	Knowledge & Logic	blind
R131	Weekday riddle from two clues	Knowledge & Logic	blind
R105	Brian's Brain	Code	blind
R104	Abelian sandpile toppling	Code	blind
R103	Wireworld with a clock and a diode	Code	blind
R102	Rule 110 growing line by line	Code	blind
R101	Langton's ant building a highway	Code	blind
R130	Radix sort ordering by one digit at a time	Code	blind
R129	Heap sort sifting the maximum to the back	Code	blind
R128	Merge sort weaving runs together	Code	blind
R127	Quicksort partitioning around glowing pivots	Code	blind
R126	Bubble sort sweeping bars into order	Code	blind
R125	Voronoi diagram growing by multi-source flood fill	Code	blind
R124	Dijkstra's shortest path over weighted terrain	Code	blind
R123	Breadth-first flood fill spreading through a region	Code	blind
R122	Randomized maze generation via recursive backtracking	Code	blind
R121	A* pathfinding search across an obstacle grid	Code	blind
R120	Scanline fill of a self-intersecting polygon	Code	blind
R119	Floyd-Steinberg dithering a gradient	Code	blind
R118	Sutherland-Hodgman polygon clipping	Code	blind
R117	De Casteljau construction of a Bezier curve	Code	blind
R116	Bresenham circles and lines, pixel by pixel	Code	blind
R115	Crowd leaving through a bottleneck	Code	blind
R114	Traffic jam from nothing	Code	blind
R113	Predator and prey on a shared field	Code	blind
R112	Physarum slime-mould network	Code	blind
R111	Ant colony laying pheromone trails	Code	blind
R110	Penrose-style aperiodic tiling growing	Code	blind
R109	Moire from two rotating rasters	Code	blind
R108	Interlocking star pattern on a grid	Code	blind
R107	Hilbert curve drawing itself	Code	blind
R106	Truchet tiles rearranging	Code	blind
R100	Sankey flow with travelling particles	Code	blind
R99	Three sorting algorithms racing	Code	blind
R98	Squarified treemap of changing data	Code	blind
R97	Circle packing without overlap	Code	blind
R96	Force-directed graph settling	Code	blind
R95	Shaded sphere by painter's algorithm	Code	blind
R94	Raycast maze walkthrough	Code	blind
R93	Rotating wireframe height field	Code	blind
R92	Rotating wireframe solids with depth sorting	Code	blind
R91	Starfield tunnel in perspective	Code	blind
R90	Poisson-disk sampling, Bridson's algorithm	Code	blind
R89	Marching squares on a moving field	Code	blind
R88	L-system tree growing in wind	Code	blind
R87	Gray-Scott reaction-diffusion	Code	blind
R86	Scrolling terrain from value noise	Code	blind
R85	Travelling salesman improved by 2-opt	Code	blind
R84	Voronoi cells with moving seeds	Code	blind
R83	Fourier epicycles tracing a shape	Code	blind
R82	Conway's Game of Life with seeded patterns	Code	blind
R81	Flocking birds with three rules	Code	blind
R80	Aggregation by continuous drift	Code	blind
R79	Aggregation on a lattice	Code	blind
R78	Crowd leaving a room by steering	Code	blind
R77	Crowd leaving a room, cell by cell	Code	blind
R76	Falling sand as free particles	Code	blind
R75	Falling sand on a cell grid	Code	blind
R74	Quadtree collision detection	Code	blind
R73	KMP string search with prefix table	Code	blind
R72	Maze generation by recursive backtracking	Code	blind
R71	Flood fill on a pixel grid	Code	blind
R70	Binary search, step by step	Code	blind
R69	Replanning when the map changes	Code	blind
R68	Edit-distance DP matrix with backtrace	Code	blind
R67	Sieve of Eratosthenes on a number grid	Code	blind
R66	AVL tree rebalancing with rotations	Code	blind
R65	Delaunay triangulation, triangle by triangle	Code	blind
R64	N-Queens solved by backtracking	Code	blind
R63	Kruskal's MST with union-find	Code	blind
R62	Huffman tree from letter frequencies	Code	blind
R61	Convex hull built step by step	Code	blind
R60	Dijkstra vs A* on a weighted grid	Code	blind
R59	Billiard break with spin and friction	Physics Lab	blind
R58	2D wave propagation with reflection	Physics Lab	blind
R57	Hanging rope with Verlet integration	Physics Lab	blind
R56	Double-slit interference, particle by particle	Physics Lab	blind
R55	The double-slit experiment and wave-particle duality (conceptual)	Physics	blind
R54	Time dilation: why do moving clocks run slow? (conceptual)	Physics	blind
R53	Why is the sky blue? (conceptual)	Physics	blind
R52	Fermi estimate: piano tuners in a 10-million-person city	Physics	blind
R51	Heat required to warm water (specific heat capacity)	Physics	blind
R50	Thin lens equation: image distance and magnification	Physics	blind
R49	Ohm's law: resistance and power dissipation	Physics	blind
R48	Low-orbit satellite: speed and period (orbital estimate)	Physics	blind
R47	Unit conversion: km/h to m/s and mph	Physics	blind
R46	Energy conservation on a frictionless incline	Physics	blind
R42	All but nine (word-problem trap)	Knowledge & Logic	blind
R43	Five machines (rate-reasoning trap)	Knowledge & Logic	blind
R44	Appointment extraction to strict JSON	Structure & Data	blind
R45	Idiom-dense German to natural English	Language & Translation	blind
R41	Prose → valid JSON	Structure & Data	blind
R40	isPalindrome (ignore case & non-alphanumerics)	Code	blind
R39	40-word product blurb with 3 required words	Creative & Marketing	blind
R38	Register shift: formal → casual (German)	Language & Translation	blind
R37	False friend: “control” EN→DE	Language & Translation	blind
R36	Continue the sequence (doubling gaps)	Knowledge & Logic	blind
R35	Minutes in a (non-leap) year — show the math	Knowledge & Logic	blind
R34	Constrained taglines (word ban + length cap)	Creative & Marketing	blind
R33	Idioms that break literal translation (EN→DE)	Language & Translation	blind
R32	Four houses, four clues (constraint puzzle)	Knowledge & Logic	blind
R31	Bat and ball (the classic reasoning trap)	Knowledge & Logic	blind
Classic rounds · R1–R5 (original 2026-07-17 write-up, editorial scores)
R5	Creative coding: an animated solar system	Physics Lab	archive
R4	Sorting race: bubble vs insertion vs quicksort	Code	archive
R3	Bouncing-ball physics: gravity, friction, elastic collisions	Physics Lab	archive
R2	Photorealistic hero-image prompt engineering	Image Models	archive
R1	Constrained product copy under a strict word count	Creative & Marketing	archive
Archive rounds · R6–R30 (names shown, pre-blind layout)
R30	Spring-mass chain	Physics Lab	archive
R29	Orbital mechanics	Physics Lab	archive
R28	Particle fountain	Physics Lab	archive
R27	Double-pendulum chaos	Physics Lab	archive
R26	Projectile ballistics	Physics Lab	archive
R25	Sorting race quick/merge/heap	Code	archive
R24	Maze race BFS/DFS/A*	Code	archive
R23	Pendulum wave	Physics Lab	archive
R22	Hexagon balls	Physics Lab	archive
R21	Sorting race bubble/insertion/selection	Code	archive
R20	JSON repair	Structure & Data	archive
R19	Table QA	Structure & Data	archive
R18	German registers & tone	Language & Translation	archive
R17	Logic riddle	Knowledge & Logic	archive
R16	Explain it to a child (German)	Language & Translation	archive
R15	CSS spinner	Code	archive
R14	Product names	Creative & Marketing	archive
R13	Image prompts	Creative & Marketing	archive
R11	Micro-copy & slogans	Creative & Marketing	archive
R10	Facts quiz	Knowledge & Logic	archive
R9	One-liner code	Code	archive
R8	Translation DE↔EN	Language & Translation	archive
R7	Sentiment classification	Structure & Data	archive
R6	Extraction → JSON	Structure & Data	archive

R1–R5 are the five classic long-form rounds (writing, image, physics, algorithms, creative coding) from the original 2026-07-17 one-pager, now published as round pages. Their answers and verdicts are unchanged from that write-up; the source page is preserved unmodified as an archival reference.

Model roster participation R1–R180

Participation covers every published round in which a model appears. "Answered" counts any round with a usable answer; only true DNFs (no answer reached the harness, or a crash) count against a model. Averages are drawn from scored rounds only — the count in brackets shows how thin or full that base is.

Model	Type	Answered	DNF	Avg score (scored)	Median latency
Local (Brain cluster)
gemma4:26b	Local	59/175	116 DNF	9.30 (5)	153.2 s
qwen3-coder:30b	Local	177/179	2 DNF	4.63 (19)	15 s
qwen3.5:9b	Local	176/179	3 DNF	4.50 (19)	14 s
qwen3:14b	Local	178/179	1 DNF	4.37 (19)	14.6 s
mistral-small:24b	Local	162/174	12 DNF	4.07 (14)	12.1 s
deepseek-r1:14b	Local	170/174	4 DNF	2.07 (14)	18.4 s
llama3.1:8b	Local	164/174	10 DNF	2.00 (14)	1.3 s
qwen3:8b	Local	39/175	136 DNF	DNF	—
deepseek-r1:32b	Local	146/150	4 DNF	DNF	—
qwq:32b	Local	133/150	17 DNF	DNF	—
command-r:35b	Local	133/150	17 DNF	DNF	—
Cloud
Claude Sonnet 5	Cloud	29/29	0	8.70 (19)	—
Claude Fable 5	Cloud	29/29	0	8.58 (19)	—
GPT-5.5 · Codex	Cloud	25/29	4 DNF	8.12 (19)	20.6 s
Claude Opus 4.8	Cloud	28/29	1 DNF	7.91 (19)	—
Claude Haiku 4.5	Cloud	24/24	0	6.18 (14)	—
Frontier reference — outside the blind field
Claude Opus 5	Cloud	179 (R1–R180)	—	not scored	—
GPT-5.5 (Codex)	Cloud	140 (R31–R180)	—	not scored	—
Claude Sonnet 5	Cloud	150 (R31–R180)	—	not scored	—
Claude Fable 5	Cloud	111 (R31–R180)	—	not scored	—
Claude Haiku 4.5	Cloud	149 (R31–R180)	—	not scored	—

Frontier references answered the identical prompts but never competed in the blind field: they are deliberately unscored and carry no latency, because they were collected through a different harness. Their answers sit in the reveal section of each round.

All rounds newest first

Categories

Model roster participation R1–R180