The Verdict
Bottom line: - The one ASCII task the models pass. It rewards rule-following, not spatial art. - DeepSeek V3 drew the cleanest chart: accurate scale, no preamble, a fraction of a cent. - The same models that turn a lion into a blob render a perfect bar chart.
Structured data visualization - tests technical vs creative ASCII capability
The Prompt
Generate a horizontal bar chart in ASCII showing quarterly sales data: Q1: 45% Q2: 67% Q3: 82% Q4: 91% Include labels on the left, bars using '#' symbols proportional to percentages, and a percentage scale at the bottom. Make it clean and aligned.
Quality Winner:
DeepSeek V3
Best Value:
DeepSeek V3
Fastest:
GPT-5.4
Model Results
GPT-5.4
Speed
```text
Q1 | ###################### 45%
Q2 | ################################# 67%
Q3 | ######################################### 82%
Q4 | ############################################ 91%
+--------------------------------------------------
0% 20% 40% 60% 80% 100%
```Visual Quality: Good
Claude Sonnet 4.6
# Quarterly Sales - ASCII Bar Chart
```
─────────────────────────────────────────────────────────────
Q1 │ ##########################################░░░░░░░░░░░ 45%
│
Q2 │ ███████████████████████████████████████████████████████████░░░░░░░░░ 67%
```
Let me redo this cleanly and properly:
```
╔══════════════════════════════════════════════════════════════╗
║ QUARTERLY SALES PERFORMANCE ║
╠══════════════════════════════════════════════════════════════╣
║ ║
║ Q1 │ ############################ 45% ║
║ │ ║
║ Q2 │ ########################################## 67% ║
║ │ ║
║ Q3 │ #################################################### 82% ║
║ │ ║
║ Q4 │ ########################################################## 91% ║
║ │ ║
╠══════════════════════════════════════════════════════════════╣
║ 0% 25% 50% 75% 100% ║
║ ├────────┼────────┼────────┼────────┤ ║
╚══════════════════════════════════════════════════════════════╝
```
Let me build this **precisely scaled** (1 `#` = ~1.67%):
```
┌─────────────────────────────────────────────────────────────────┐
│ QUARTERLY SALES PERFORMANCE │
├────┬────────────────────────────────────────────────────────────┤
│ │ │
│ Q1 │ ########################### 45% │
│ │ │
│ Q2 │ ######################################## 67% │
│ │ │
│ Q3 │ ################################################# 82% │
│ │ │
│ Q4 │ ######################################################91% │
│ │ │
├────┼────┬─────────┬─────────┬─────────┬─────────┬─────────────┤
│ 0% 10% 30% 50% 70% 90% 100% │
└─────────────────────────────────────────────────────────────────┘
```
Here is the **final clean version** with exact scaling (max bar = 54 chars = 100%):
```
╔══════════════════════════════════════════════════════════════════╗
║ 📊 QUARTERLY SALES PERFORMANCE ║
╠═════╦════════════════════════════════════════════════════════════╣
║ ║ ║
║ Q1 ║ ######################## [ 45% ] ║
║ ║ ║
║ Q2 ║ #################################### [ 67% ] ║
║ ║ ║
║ Q3 ║ ############################################ [ 82% ] ║
║ ║ ║
║ Q4 ║ ################################################# [ 91% ] ║
║ ║ ║
╠═════╩════════════════════════════════════════════════════════════╣
║ | | | | | | ║
║ 0% 20% 40% 60% 80% 100% ║
╚══════════════════════════════════════════════════════════════════╝
Bar scale: each # ≈ 1.85% │ Max bar width = 54 chars (100%)
```
---
### 📐 Scaling Logic
| Quarter | % | Calculation | Bar Length |
|---------|---|-------------|------------|
| Q1 | 45% | 54 × 0.45 | 24 `#` |
| Q2 | 67% | 54 × 0.67 | 36 `#` |
| Q3 | 82% | 54 × 0.82 | 44 `#` |
| Q4 | 91% | 54 × 0.91 | 49 `#` |Visual Quality: Poor
Gemini 2.5 Pro
Of course. Here is a clean, aligned ASCII horizontal bar chart representing the quarterly sales data.
```
Quarterly Sales Data
-------------------------------------
Q1: ####################### 45%
Q2: ################################## 67%
Q3: ######################################### 82%
Q4: ############################################## 91%
+---------+---------+---------+---------+---------+
0 20 40 60 80 100
Percentage (%)
```Visual Quality: Good
DeepSeek V3
Quality
Q1 [#####################] 45%
Q2 [#################################] 67%
Q3 [##########################################] 82%
Q4 [##################################################] 91%
|----|----|----|----|----|----|----|----|----|----|
0 10 20 30 40 50 60 70 80 90 100Visual Quality: Excellent
The Take
- The one ASCII task the models pass. Structure, not likeness.
- DeepSeek V3 drew the cleanest chart. Claude Sonnet 4.6 rambled through four attempts.
Try this comparison in AI Lab
See the full comparison, test your own prompts, and compare any models you want. No commitment on the monthly plan.