Qwen 3.8 Max

Alibaba. Run in OpenCode.

12 runs on 12 prompts · 50% finished

Results

The best run of each of its 12 tests first, then every other run. Filter them in the explorer

6 finished runs

Live demo
Marathon results
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Live demo
Sorting race
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Live demo
World weather stations
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Live demo
Football league season
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Live scene
Impressionist world
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Live scene
Mobile touch arcade game
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)

Folding road
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Failed · Did not finish
GPU fluid sim
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Failed · Did not finish
Js13k style shader game
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Failed · Did not finish
Game battleship
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Failed · Did not finish
Pathfinding race
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Failed · Did not finish
Island stream diorama
Qwen 3.8 Max
OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)
Failed · Stopped by us

Numbers: finish rate, cost, wall time, draft scores

Runs: 12; on 12 prompts
Finished: 50%; 6 of 12
Median cost, finished run: $1.79; n=6
Median wall time, finished run: 21m 13s; n=6
Recorded spend: $17.00; $4.24 on unfinished runs

From the numbers

Set against the other models with at least 3 runs (top or bottom third), and its draft-score ranks within single prompts. Nothing here is averaged across prompts.

Strengths

Nothing stands out with enough runs to say so.

Weaknesses

6 of 12 runs finished: 20th of 26 models with at least 3 (finish rate)
21m 13s (n=6): 22nd of 24 models with at least 3 (median wall time of a finished run)
$1.79 (n=6): 13th of 16 models with at least 3 (median cost of a finished run)
5 runs ended as the model's own failure (DNF, class MODEL)
Other unfinished runs, not counted against the model: 1 TIMEOUT

Best and hardest run

Best run

Picked as its fastest finished run (no scored result yet).

Live page

Qwen 3.8 Max, OpenCode, algoviz_sorting_race

OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)

11m 20s

$0.81 reported by OpenCode

OK

Unscored: no rubric for this prompt

Hardest run

Picked as its costliest run that did not finish.

Did not finish · re-run queued

Qwen 3.8 Max, OpenCode, threejs_island_stream_diorama

OpenCode 1.18.35
skills on · OpenCode 1.18.35 + skills on (54: 33 video + 21 web/Three.js)

Stopped by us

45m 01s $2.75 reported by OpenCode

Draft scores by prompt

No scored result yet (only prompts with a published rubric are scored).

By harness


OpenCode	12	6	50%	$1.79 n=6	21m 13s n=6

All 12 tests and experiment collections

Experiment collections: HTML Files (7) · Three.js (7)