TASK 12 / HTML · V1 · Utility

Calculator App

计算器应用

4 runs · 12.A.01 · HTML

Compare works →
TASK 12 · 12.A.01 · HTMLCalculator App
Open independent copy ↗
Calculator App

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 working calculator app.
Choose the technology yourself.
Run it.
Test it.
Fix any issues you find.
Do not ask me questions unless you are genuinely blocked.
TRANSLATION

Chinese translation

创建一个可工作的计算器应用。技术由你自行选择。运行它、测试它,并修复发现的所有问题。除非确实受阻,否则不要向我提问。

Expected outcome

The deliverable should be a calculator application that actually runs and has been tested and repaired, using a technology stack chosen by the implementer. Minimum capabilities include digit and decimal entry, the four basic operations, clear, backspace, sign toggle, and equals. A strong implementation should also handle precedence, chained calculations, keyboard input, error states, and number formatting. The calculation core should be deterministic; division by zero, malformed expressions, overflow, and floating-point display noise should be handled intelligibly without crashing the UI. The handoff should include run instructions, a test method, and evidence that discovered defects were fixed.

Additional verification guidance

  • Run the real app and use both buttons and keyboard for basic, chained, decimal, negative, and precedence-sensitive calculations against known results.
  • Cover cases such as `0.1 + 0.2`, division by zero, repeated decimal points, leading operators, long input, overflow, and recovery after an error.
  • Check state-machine consistency for clear, backspace, sign toggle, repeated equals, and consecutive operators, with no console exceptions.
  • Review keypad layout, focus, readability, and touch targets on desktop and narrow screens; run automated tests and a full regression after fixes.
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-12-r1Muse Spark 1.3run-muse-spark-1-3-task-12-r1GPT-5.6 Solrun-gpt-5-6-sol-task-12-r1K3run-k3-task-12-r1
Runtime6:202:006:044:05
API calls89181612
Tool calls9021811
Failures / retries6001
Token throughput7.5M316K609.3K382.5K
AI review98/10078/10078/10089/100
RUN RECORDS

Run records

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

12.A.01 · run-deepseek-v4-1-flash-exp-0910-task-12-r1DeepSeek-V4.1-Flash● Archived
Runtime
6:20
Token throughput
7.5M
API calls
89
View archive record →
12.B.01 · run-muse-spark-1-3-task-12-r1Muse Spark 1.3● Archived
Runtime
2:00
Token throughput
316K
API calls
18
View archive record →
12.C.01 · run-gpt-5-6-sol-task-12-r1GPT-5.6 Sol● Archived
Runtime
6:04
Token throughput
609.3K
API calls
16
View archive record →
RELATED TASKS

Related tasks

View all tasks →