OpenAI Usage Dashboard — May 2026
Professional Usage Dashboard

OpenAI API Usage
May 2026

Executive summary of model requests, token consumption, cache usage, model mix, API-key distribution, and daily spikes. Sensitive project/user/API identifiers are masked in the dashboard.

Text-only usage 14 models 5 API keys 21/30 active days
Coverage
01 May – 30 May 2026
Rows Analyzed
64 usage rows
Zero-Usage Days
9
Data Quality
Clean
Total model requests
1,430
Across all models
Total tokens
3.84M
3,837,137 exact tokens
Input tokens
3.60M
Prompt/context consumption
Output tokens
240.1K
6.7% of input volume
Cached input tokens
1.20M
33.4% input cache ratio
Uncached input tokens
2.40M
Primary optimization area
Avg tokens/request
2,683
Blended across models
Peak day
12 May
2.56M tokens

Executive Insights

High-level interpretation for leadership and cost-control review
Usage is highly concentrated.12 May 2026 alone consumed 2.56M tokens, representing 66.6% of total usage. This should be reviewed as a spike day.
gpt-5.5-2026-04-23 dominates consumption.This model used 2.59M tokens, or 67.5% of total tokens, despite only 293 requests.
Routine traffic appears model-efficient.gpt-4.1-mini-2025-04-14 had the highest request count with 465 requests, but a smaller token share than the largest model.
Prompt caching is helping.1.20M cached input tokens reduced repeated context load, with an overall cache ratio of 33.4%.
All usage is text-based.Audio and image token columns are zero, so this dashboard reflects text/token workloads only.
There are idle days.9 days in the range had no model usage. This is useful for separating real demand from event/demo spikes.

Daily Token Trend

Total input + output tokens
0 639.2K 1.28M 1.92M 2.56M 01 May — 62,407 tokens, 50 requests 01 02 May — 134,750 tokens, 312 requests 03 May — 0 tokens, 0 requests 04 May — 5,711 tokens, 7 requests 05 May — 0 tokens, 0 requests 06 May — 0 tokens, 0 requests 06 07 May — 0 tokens, 0 requests 08 May — 0 tokens, 0 requests 09 May — 54,101 tokens, 17 requests 10 May — 451 tokens, 2 requests 11 May — 0 tokens, 0 requests 11 12 May — 2,556,796 tokens, 94 requests 13 May — 80,756 tokens, 73 requests 14 May — 0 tokens, 0 requests 15 May — 8,124 tokens, 10 requests 16 May — 59,533 tokens, 76 requests 16 17 May — 3,506 tokens, 4 requests 18 May — 20,088 tokens, 22 requests 19 May — 167,468 tokens, 230 requests 20 May — 170,395 tokens, 208 requests 21 May — 384,125 tokens, 254 requests 21 22 May — 0 tokens, 0 requests 23 May — 18,765 tokens, 7 requests 24 May — 19,714 tokens, 20 requests 25 May — 2,894 tokens, 4 requests 26 May — 2,244 tokens, 3 requests 26 27 May — 0 tokens, 0 requests 28 May — 200 tokens, 1 requests 29 May — 55,018 tokens, 10 requests 30 May — 30,091 tokens, 26 requests 30 May 2026 daily total tokens. Hover bars for details.

Daily Request Trend

Model calls per day
0 104 208 312 01 May — 50 requests 01 02 May — 312 requests 03 May — 0 requests 04 May — 7 requests 05 May — 0 requests 06 May — 0 requests 06 07 May — 0 requests 08 May — 0 requests 09 May — 17 requests 10 May — 2 requests 11 May — 0 requests 11 12 May — 94 requests 13 May — 73 requests 14 May — 0 requests 15 May — 10 requests 16 May — 76 requests 16 17 May — 4 requests 18 May — 22 requests 19 May — 230 requests 20 May — 208 requests 21 May — 254 requests 21 22 May — 0 requests 23 May — 7 requests 24 May — 20 requests 25 May — 4 requests 26 May — 3 requests 26 27 May — 0 requests 28 May — 1 requests 29 May — 10 requests 30 May — 26 requests 30

Model Mix

Sorted by total tokens
Model Requests Input Output Total Token Share Cache Ratio
gpt-5.5-2026-04-23 293 2.52M 68.4K 2.59M
67.5% of total
31.5%
gpt-5.4-mini-2026-03-17 74 422.3K 27.7K 450.0K
11.7% of total
67.0%
gpt-4.1-mini-2025-04-14 465 310.6K 64.0K 374.7K
9.8% of total
15.9%
gpt-5-mini-2025-08-07 111 107.5K 31.6K 139.1K
3.6% of total
22.9%
gpt-5.4-2026-03-05 56 85.8K 7.2K 93.0K
2.4% of total
53.4%
gpt-5.1-2025-11-13 68 42.1K 6.4K 48.5K
1.3% of total
0.0%
gpt-5.4-nano-2026-03-17 6 46.3K 2.1K 48.4K
1.3% of total
0.0%
gpt-4o-mini-2024-07-18 168 42.5K 4.1K 46.6K
1.2% of total
8.1%
gpt-4.1-2025-04-14 62 2.8K 20.9K 23.7K
0.6% of total
0.0%
gpt-4.1-nano-2025-04-14 55 6.3K 3.6K 9.8K
0.3% of total
0.0%
gpt-3.5-turbo-0125 54 7.6K 637 8.2K
0.2% of total
0.0%
gpt-5-pro-2025-10-06 1 83 2.0K 2.1K
0.1% of total
0.0%
gpt-4o-2024-08-06 16 562 1.2K 1.8K
0.0% of total
0.0%
gpt-5-chat-latest 1 611 330 941
0.0% of total
0.0%

Top Usage Days

Highest token days
Date Requests Input Output Total Cache
12 May 2026 94 2.53M 26.6K 2.56M 35.4%
21 May 2026 254 342.8K 41.3K 384.1K 46.4%
20 May 2026 208 141.5K 28.9K 170.4K 7.1%
19 May 2026 230 114.8K 52.7K 167.5K 30.7%
02 May 2026 312 112.9K 21.8K 134.8K 9.0%
13 May 2026 73 60.6K 20.2K 80.8K 27.7%
01 May 2026 50 53.0K 9.4K 62.4K 28.7%
16 May 2026 76 47.1K 12.5K 59.5K 73.2%
29 May 2026 10 52.0K 3.0K 55.0K 30.0%
09 May 2026 17 49.8K 4.3K 54.1K 0.0%

API Key Distribution

Masked key IDs
API Key Requests Tokens Share Cache
key_Vbg4eB…DiAL 1,404 3.73M
97.3% share
33.6%
key_4veODg…PSiW 10 55.0K
1.4% share
30.0%
key_tabJc4…uUqk 9 26.5K
0.7% share
35.8%
key_y7rbJP…9xaQ 6 14.9K
0.4% share
0.0%
key_b60njZ…r50x 1 7.7K
0.2% share
0.0%

Recommendations

Practical next actions
1. Investigate spike

Review the workflows that ran on 12 May 2026. The token spike is large enough to deserve a separate explanation: demo, batch run, long-context test, loop, or accidental repeated call.

2. Route by task

Use smaller/mini/nano models for summaries, classification, formatting, and simple Q&A. Reserve the largest model for reasoning-heavy, high-value, or client-facing outputs.

3. Expand caching

Because one-third of input tokens were cached, standardize reusable system prompts, fixed instructions, and static context blocks to improve repeatability and reduce uncached volume.

4. Improve logging

Add metadata per request: product, course/event, workflow name, environment, user group, and feature. This will make the next dashboard more business-friendly and easier to map to ROI.

5. Separate keys

Create separate API keys by environment or product, such as demo, bootcamp, AI Guild, internal tools, and client projects. This will make governance and budget tracking cleaner.

6. Add cost layer

This dashboard intentionally avoids cost estimates because model pricing was not included in the source data. Add a pricing table later to calculate exact blended cost by model.

Daily Detail

Includes zero-usage days for context
Date Requests Input Tokens Output Tokens Cached Tokens Total Tokens
01 May 2026 50 53,022 9,385 15,232 62,407
02 May 2026 312 112,913 21,837 10,112 134,750
03 May 2026 0 0 0 0 0
04 May 2026 7 4,991 720 1,152 5,711
05 May 2026 0 0 0 0 0
06 May 2026 0 0 0 0 0
07 May 2026 0 0 0 0 0
08 May 2026 0 0 0 0 0
09 May 2026 17 49,763 4,338 0 54,101
10 May 2026 2 315 136 0 451
11 May 2026 0 0 0 0 0
12 May 2026 94 2,530,147 26,649 895,488 2,556,796
13 May 2026 73 60,586 20,170 16,768 80,756
14 May 2026 0 0 0 0 0
15 May 2026 10 6,550 1,574 0 8,124
16 May 2026 76 47,055 12,478 34,432 59,533
17 May 2026 4 126 3,380 0 3,506
18 May 2026 22 17,653 2,435 3,072 20,088
19 May 2026 230 114,792 52,676 35,200 167,468
20 May 2026 208 141,455 28,940 10,112 170,395
21 May 2026 254 342,840 41,285 158,976 384,125
22 May 2026 0 0 0 0 0
23 May 2026 7 14,302 4,463 0 18,765
24 May 2026 20 14,763 4,951 1,024 19,714
25 May 2026 4 2,630 264 0 2,894
26 May 2026 3 2,021 223 0 2,244
27 May 2026 0 0 0 0 0
28 May 2026 1 145 55 0 200
29 May 2026 10 52,045 2,973 15,616 55,018
30 May 2026 26 28,904 1,187 3,456 30,091

Leave a Reply

Your email address will not be published. Required fields are marked *