Two quests.
Ten minutes.

Median image

Who made it?

Selected native setup per model, median time, setup-specific finish counts and API-price equivalent
Model / setupMedian · 10 min capped05 min10 minFinishesAvg. API USDRecording
GPT-6.2 SolCodex CLI 0.167.1 · ChatGPT subscription · Standard · Medium reasoning · Setup note17 Nov · three attempts
5:07n=3
3 / 3$0.47
Fable 5.2Claude Code 2.1.321 · Anthropic / Claude Pro subscription · Standard · Medium reasoning · clock · keep playing10 Nov · three attempts
6:22n=3
3 / 3≥$2.26
GPT-6.2 AstraCodex CLI 0.167.1 · ChatGPT subscription · Standard · Low reasoning · Setup note17 Nov · three attempts
6:28n=3
3 / 3$1.18
Opus 5.5Claude Code 2.1.319 · Anthropic · Standard · High reasoning · clock · keep playing · 2× minimapClaude Code settings · 8 Nov
7:13n=3
3 / 3≥$2.12
GPT-6.1 SolCodex · OpenAI · Standard · Medium reasoningSol reasoning pilot
7:26n=2
2 / 2≥$0.50
GPT-6 AstraCodex · OpenAI · Standard · Low reasoningOriginal comparison
≥10:00n=3 · unfinished runs
1 / 3≥$3.24
GPT-6.1 TerraCodex · OpenAI · Fast requested · Low reasoningOriginal comparison
≥10:00n=3 · unfinished runs · quest 1 at 2:30
0 / 3≥$0.93
Claude Sonnet 5.5Claude Code 2.1.316 · Anthropic · Standard · Low reasoningOriginal comparison
≥10:00n=1 · unfinished runs · quest 1 at 3:40
0 / 1≥$1.34
Gemini 4 CarbonGemini CLI 0.68.0 · Gemini API (prepaid credit) · Standard · Medium reasoning11 Nov · three attempts
≥10:00n=3 · unfinished runs · quest 1 at 5:33
0 / 3$0.42

Codex, Claude Code and Gemini CLI only. Setups differ; finish counts apply to the displayed setup.

Median includes unfinished runs at the ten-minute limit; ≥ marks a lower bound. Price averages all scored attempts. Cost ≥ means partial usage; API equivalents are not cash charges.

Historical human reference · n = 12:27