First passes
Which checks fell first, and which no model has passed yet.
Each row is one check, with the first run that passed it and how many of the models tried have passed it at least once. Only runs on the task’s current prompt and checks count. The dates are when I ran each model, so the order says as much about which task I ran first as about the models.
24 models, 65 runs, prompt v2, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| All seven parts | eye | 2026-09-22, Pareto | 16 of 24 |
| Stack order | measured | 2026-09-22, Pareto | 13 of 24 |
| Filter fits the glass | measured | 2026-09-22, Pareto | 17 of 24 |
| One assembly axis | eye | 2026-09-22, Pareto | 21 of 24 |
| Labels match | eye | 2026-09-22, Pareto | 22 of 24 |
5 models, 15 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| All five parts | eye | 2026-10-05, GPT-6 Luna | 4 of 5 |
| Stack order | measured | 2026-10-05, GPT-6 Luna | 4 of 5 |
| Parts fit | measured | 2026-10-07, GPT-6.1 Sol | 3 of 5 |
| One assembly axis | eye | 2026-10-05, GPT-6 Luna | 5 of 5 |
| Labels match | eye | 2026-10-05, GPT-6 Luna | 5 of 5 |
4 models, 12 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| All three parts | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
| Stack order | measured | 2026-10-05, GPT-6 Luna | 3 of 4 |
| Parts fit | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
| One assembly axis | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
| Labels match | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
8 models, 22 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| All five parts | eye | 2026-10-05, GPT-6 Luna | 6 of 8 |
| Stack order | measured | 2026-10-05, GPT-6 Luna | 6 of 8 |
| Parts fit | eye | 2026-10-05, GPT-6 Luna | 7 of 8 |
| Two assembly axes | eye | 2026-10-05, GPT-6 Luna | 8 of 8 |
| Labels match | eye | 2026-10-05, GPT-6 Luna | 8 of 8 |
4 models, 12 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| All eight parts | eye | 2026-10-07, GPT-6.1 Sol | 2 of 4 |
| Stack order | measured | 2026-10-05, GPT-6 Luna | 3 of 4 |
| Parts fit | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
| One assembly axis | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
| Labels match | eye | 2026-10-05, GPT-6 Luna | 4 of 4 |
3 models, 6 runs, prompt v2, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| On real streets | measured | 2026-09-17, Union Alpha | 3 of 3 |
| One continuous route | measured | 2026-09-17, Union Alpha | 3 of 3 |
| Inside the grid | measured | 2026-09-17, Union Alpha | 3 of 3 |
| Metadata true | eye | 2026-09-17, Union Alpha | 3 of 3 |
| Reads as COFFEE | eye | 2026-09-17, Union Alpha | 2 of 3 |
2 models, 9 runs, prompt v1, checks c1, harness h1, saturated 2026-09-21
| check | decided by | first passed | models that passed |
|---|---|---|---|
| C8H10N4O2 | measured | 2026-09-14, Qwen3.8 Max | 2 of 2 |
| Fused purine core | measured | 2026-09-14, Qwen3.8 Max | 2 of 2 |
| Methyls on N1, N3, N7 | measured | 2026-09-14, Qwen3.8 Max | 2 of 2 |
| Carbonyls at C2 and C6 | measured | 2026-09-14, Qwen3.8 Max | 2 of 2 |
| Flat rings, real bond lengths | measured | 2026-09-14, Qwen3.8 Max | 2 of 2 |
2 models, 5 runs, prompt v2, checks c2, harness h1, saturated 2026-09-21
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Complete analogue | measured | 2026-09-21, Astra | 2 of 2 |
| Real hysteresis | measured | 2026-09-21, Astra | 2 of 2 |
| Starts cold | measured | 2026-09-21, Astra | 2 of 2 |
| Holds 92–96 °C | measured | 2026-09-21, Astra | 2 of 2 |
| Writes the .txt | measured | 2026-09-21, Astra | 2 of 2 |
2 models, 5 runs, prompt v2, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| All five objects | eye | 2026-09-21, DeepSeek 4.1 Flash | 2 of 2 |
| Primitives only | measured | 2026-09-21, DeepSeek 4.1 Flash | 2 of 2 |
| Spoon reads as a spoon | eye | 2026-09-21, Qwen3.8 Flash | 1 of 2 |
| Croissant reads as a croissant | eye | 2026-09-21, Qwen3.8 Flash | 2 of 2 |
| Consistent scale | measured | 2026-09-21, DeepSeek 4.1 Flash | 2 of 2 |
Caffeine with three CF3 groups
2 models, 6 runs, prompt v1, checks c1, harness h1, saturated 2026-10-02
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Parses | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| C8HF9N4O2 | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| Three N-CF3, no N-methyl | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| The right molecule | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
A three-set coffee Venn diagram
17 models, 61 runs, prompt v1, checks c2, harness h1, saturated 2026-10-13
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Compiles | measured | 2026-09-22, DeepSeek 4.1 Flash | 17 of 17 |
| Seven regions, three names | measured | 2026-09-22, DeepSeek 4.1 Flash | 17 of 17 |
| Every drink in its region | measured | 2026-09-22, DeepSeek 4.1 Flash | 16 of 17 |
| No label on an edge | measured | 2026-09-22, DeepSeek 4.1 Flash | 10 of 17 |
A knitting chart for a coffee-bean cup cosy
7 models, 21 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| A real chart | measured | 2026-10-11, GPT-6.1 Sol | 7 of 7 |
| Stitches wider than tall | measured | 2026-10-11, GPT-6.1 Sol | 7 of 7 |
| Bean knits true | measured | 2026-10-11, GPT-6.1 Sol | 7 of 7 |
| Has a key | measured | 2026-10-11, GPT-6.1 Sol | 7 of 7 |
A five-set Venn in circles, which cannot be drawn
6 models, 16 runs, prompt v1, checks c1, harness h1, saturated 2026-10-09
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Says circles cannot | measured | 2026-10-09, Pareto 26.10 Preview | 6 of 6 |
| Five sets kept | measured | 2026-10-09, Pareto 26.10 Preview | 6 of 6 |
| Names a fix that works | eye | 2026-10-09, Pareto 26.10 Preview | 4 of 6 |
Triangle-wave carrier for a milk frother
2 models, 6 runs, prompt v1, checks c1, harness h1, saturated 2026-10-02
| check | decided by | first passed | models that passed |
|---|---|---|---|
| DC supplies only | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| 1 kHz | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| 2 V peak to peak | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| Straight, even ramps | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
A café order that can be cancelled
2 models, 6 runs, prompt v1, checks c1, harness h1, saturated 2026-10-02
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Renders, labels valid | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| Happy path | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| No refund if unpaid | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| Refund if paid | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
| No cancel after collect | measured | 2026-09-22, DeepSeek 4.1 Flash | 2 of 2 |
A coffee shrub hedge-pruned flat on one side
21 models, 50 runs, prompt v1, checks c2, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Reads as a shrub | eye | 2026-09-29, DeepSeek 4.1 Flash | 13 of 21 |
| Flat side from the rules | eye | 2026-09-29, DeepSeek 4.1 Flash | 15 of 21 |
| Sensible size | eye | 2026-09-29, DeepSeek 4.1 Flash | 14 of 21 |
2 models, 6 runs, prompt v1, checks c1, harness h1, saturated 2026-10-02
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Two rhombi, two fills | eye | 2026-10-02, Pareto 26.10 Preview | 2 of 2 |
| No repeat | eye | 2026-10-02, Pareto 26.10 Preview | 2 of 2 |
| Fivefold, edges drawn | eye | 2026-10-02, Pareto 26.10 Preview | 2 of 2 |
Espresso Rag, with the left hand crossing over
4 models, 12 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Piano staff | measured | 2026-10-04, GPT-6.1 Sol | 4 of 4 |
| Sixteen bars | measured | 2026-10-04, GPT-6.1 Sol | 4 of 4 |
| Left hand crosses | measured | 2026-10-04, GPT-6.1 Sol | 4 of 4 |
| Right hand keeps going | measured | 2026-10-04, GPT-6.1 Sol | 4 of 4 |
| Crossing marked | measured | 2026-10-04, GPT-6.1 Sol | 2 of 4 |
| Stems apart | measured | 2026-10-04, GPT-6.1 Sol | 4 of 4 |
A family tree of twelve coffee cultivars
5 models, 15 runs, prompt v1, checks c1, harness h1, saturated 2026-10-14
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Valid Newick | measured | 2026-10-14, Gemini 4 Argon | 5 of 5 |
| Twelve cultivars | measured | 2026-10-14, Gemini 4 Argon | 5 of 5 |
| Every tip level | measured | 2026-10-14, Gemini 4 Argon | 5 of 5 |
| One split of four | measured | 2026-10-14, Gemini 4 Argon | 5 of 5 |
Beans leaving a grinder hopper, in MuJoCo
5 models, 13 runs, prompt v1, checks c1, harness h1, saturated 2026-11-06
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Runs in MuJoCo | measured | 2026-10-15, Haiku 5.5 | 5 of 5 |
| model.xml written | measured | 2026-10-15, Haiku 5.5 | 5 of 5 |
| Fifty two-part beans | measured | 2026-10-15, Haiku 5.5 | 5 of 5 |
| Gravity down | measured | 2026-10-15, Haiku 5.5 | 5 of 5 |
| Beans start inside a hollow hopper | measured | 2026-10-15, Haiku 5.5 | 4 of 5 |
| Held, then flows | model | 2026-10-15, Pareto 26.9 | 4 of 5 |
A spoon stirring in a cup, in HTML and CSS
3 models, 9 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| HTML and CSS only | measured | 2026-10-19, Claude Sonnet 5.5 | 3 of 3 |
| A CSS loop that moves something | measured | 2026-10-19, Claude Sonnet 5.5 | 3 of 3 |
| Reads as a cup seen into | eye | 2026-10-19, Claude Sonnet 5.5 | 3 of 3 |
| The spoon does the stirring | eye | 2026-10-19, Claude Sonnet 5.5 | 2 of 3 |
| Reads as inside the cup | eye | 2026-10-19, Claude Sonnet 5.5 | 2 of 3 |
| Handle stays on the bowl | eye | 2026-10-19, Claude Sonnet 5.5 | 3 of 3 |
Repair: a V60 exploded diagram with the filter under the cone
3 models, 9 runs, prompt v1, checks c1, harness h1, saturated 2026-10-19
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Repaired | measured | 2026-10-19, GLM 5.3 FlashX | 3 of 3 |
| Broke nothing else | measured | 2026-10-19, GLM 5.3 FlashX | 3 of 3 |
Repair: a stirring spoon laid across the mouth of the cup
3 models, 9 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| In the drink | eye | 2026-10-19, Claude Sonnet 5.5 | 2 of 3 |
| Kept the brief and the loop | measured | 2026-10-19, Claude Sonnet 5.5 | 3 of 3 |
Repair: a Venn diagram with a drink on a circle's edge
3 models, 9 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Repaired | measured | 2026-10-19, DeepSeek 4.1 Flash | 3 of 3 |
| Broke nothing else | measured | 2026-10-19, DeepSeek 4.1 Flash | 3 of 3 |
Repair: a knitting chart whose bean knits too long
3 models, 9 runs, prompt v1, checks c1, harness h1, saturated 2026-10-19
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Repaired | measured | 2026-10-19, Claude Haiku 5.5 | 3 of 3 |
| Broke nothing else | measured | 2026-10-19, Claude Haiku 5.5 | 3 of 3 |
Autopsy, asked if it is correct: Claude Opus 5.5's French press as a picture (control)
2 models, 7 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Invents nothing | measured | 2026-10-19, Opus 5.5 | 2 of 2 |
Autopsy, asked if it is correct: Claude Opus 5.5's French press as a picture
2 models, 8 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Finds the flip | measured | 2026-10-19, Opus 5.5 | 2 of 2 |
| Invents nothing | measured | 2026-10-19, Opus 5.5 | 2 of 2 |
A serving tray where every table gets a drink it will take
8 models, 21 runs, prompt v2, checks c1, harness h1, saturated 2026-11-08
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Serves all six | measured | 2026-11-08, Opus 5.5 | 8 of 8 |
A serving tray where four tables will only take three drinks
8 models, 24 runs, prompt v2, checks c1, harness h1, saturated 2026-11-10
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Says it cannot be done | measured | 2026-11-08, GPT-6.1 Sol | 8 of 8 |
| Reuses nothing | measured | 2026-11-08, GLM 5.3 FlashX | 6 of 8 |
A serving tray where two tables both want black coffee
8 models, 24 runs, prompt v2, checks c1, harness h1, saturated 2026-11-08
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Serves all six | measured | 2026-11-08, Kimi K3.1 Pro | 8 of 8 |
The coffee machine taken out of the café's music
5 models, 13 runs, prompt v1, checks c1, harness h1
| check | decided by | first passed | models that passed |
|---|---|---|---|
| Same file | measured | 2026-11-09, GPT-6.1 Terra | 5 of 5 |
| Hum down | measured | 2026-11-09, GPT-6.1 Terra | 3 of 5 |
| Rush down | measured | 2026-11-09, GPT-6.1 Terra | 3 of 5 |
| Music kept | measured | 2026-11-09, GPT-6.1 Terra | 3 of 5 |
| Sounds right | eye | not yet | 0 of 5 |