TASK 09 / HTML · V1 · Game

Single-File Top-Down Farming Game

单文件俯视角农场游戏

4 runs · 09.A.01 · HTML

Compare works →
TASK 09 · 09.A.01 · HTMLSingle-File Top-Down Farming Game
Open independent copy ↗
Single-File Top-Down Farming Game

Preview note: Embedded storage and downloads can be restricted by browser sandboxing; use the standalone view for full behavior.

ORIGINAL PROMPT

Original prompt

PROMPT/Phase1_TEST_PROMPTS_BILINGUAL.md · The original prompt is kept verbatim

Source ↗
ORIGINAL PROMPT

Original prompt

Create a playable top-down farming game in a single HTML file.
TRANSLATION

Chinese translation

在单个 HTML 文件中创建一个可玩的俯视角农场游戏。

Expected outcome

The deliverable should be one fully self-contained HTML file with a controllable character and a recognizable farm presented from a top-down view. It should provide at least a complete loop of preparing soil, planting, tending, growth, and harvesting. The game needs clear controls, state feedback, changing resources or time, and a goal or progression that can be sustained over multiple cycles. The map, avatar, crop stages, and interactive areas should be visually distinguishable, while collisions, boundaries, and interaction distance remain consistent. Optional money, shops, weather, stamina, or saving systems should support the core loop and maintain coherent state.

Additional verification guidance

  • Open the file offline and complete a full planting-to-harvest cycle from a new game, confirming visible state changes at every step.
  • Verify movement, map boundaries, obstacle collision, tool switching, interaction range, resource deductions, and insufficient-resource feedback.
  • Test crop stages, day or time progression, growth conditions such as watering, and correct inventory or currency updates after harvest.
  • Run prolonged and randomized input to detect loops, state corruption, major performance degradation, or console errors; verify playability across viewports.
THREE INTERPRETATIONS

Same task, three model interpretations

All three results use the same width and the same preview ratio; no model is marked as best.

COMPARISON METRICS

Comparison metrics

Cross-model comparison must note differences in harness, budget and isolation.

MetricDeepSeek-V4.1-Flashrun-deepseek-v4-1-flash-exp-0910-task-09-r1Muse Spark 1.3run-muse-spark-1-3-task-09-r1GPT-5.6 Solrun-gpt-5-6-sol-task-09-r1K3run-k3-task-09-r1
Runtime11:341:415:1011:01
API calls427415
Tool calls416314
Failures / retries2002
Token throughput3.6M113.1K129.9K699.7K
AI review88/10084/10082/10093/100
RUN RECORDS

Run records

Two runs of the same task version can be compared side by side.

09.A.01 · run-deepseek-v4-1-flash-exp-0910-task-09-r1DeepSeek-V4.1-Flash● Archived
Runtime
11:34
Token throughput
3.6M
API calls
42
View archive record →
09.B.01 · run-muse-spark-1-3-task-09-r1Muse Spark 1.3● Archived
Runtime
1:41
Token throughput
113.1K
API calls
7
View archive record →
09.C.01 · run-gpt-5-6-sol-task-09-r1GPT-5.6 Sol● Archived
Runtime
5:10
Token throughput
129.9K
API calls
4
View archive record →
RELATED TASKS

Related tasks

View all tasks →