You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Over the last 30 days of collected data, Copilot-powered agentic workflows consumed 90,369,806 tokens (~$271.11 estimated*) across 101 workflow runs spanning 50 unique workflows. Today's snapshot covers all completed Copilot runs visible in the audit log.
*Cost estimated at ~$3.00/1M tokens (claude-sonnet approximate average). Actual pricing may vary by model and context.
Key Highlights
Metric
Value
π’ Total Token Usage
90,369,806
π° Estimated Cost
~$271.11
π Total Workflow Runs
101
ποΈ Unique Workflows
50
β‘ Avg Tokens / Run
~895K
π Highest Single Consumer
Daily Syntax Error Quality Check (5.6M tokens)
π Most Active Workflow
Issue Monster (18 runs)
π Trend vs Last Report (2026-02-20)
-46% tokens, -48% runs
π Token Usage & Cost Trends
Historical Token Consumption
Token usage shows a downward trend from the February peaks (~237M tokens on 2026-02-11) to the current 90M. Today's run count of 101 is the second lowest on record, down from a peak of 378 runs.
Estimated Daily Cost Trends
β οΈNote: Cost values in earlier history entries used different pricing models, resulting in visible inconsistencies. Today's estimate ($271.11) uses the current $3.00/1M token model consistently.
π Top 10 Workflows by Token Consumption
Rank
Workflow
Total Tokens
Est. Cost
Runs
Avg Tokens/Run
Avg Cost/Run
Avg Turns
Avg Duration
1
Daily Syntax Error Quality Check
5,624,276
$16.87
1
5,624,276
$16.87
1.0
18.7m
2
Daily Community Attribution Updater
5,615,968
$16.85
1
5,615,968
$16.85
116.0
19.2m
3
jsweep - JavaScript Unbloater
5,539,793
$16.62
1
5,539,793
$16.62
22.8m
22.8m
4
Contribution Check
5,411,039
$16.23
6
901,839
$2.71
11.0
7.9m
5
Auto-Triage Issues
5,409,071
$16.23
16
338,066
$1.01
6.2
5.8m
6
Dead Code Removal Agent
4,682,216
$14.05
1
4,682,216
$14.05
56.0
18.5m
7
Issue Monster
4,468,561
$13.41
18
248,253
$0.74
4.7
6.0m
8
Smoke Multi PR
4,165,940
$12.50
1
4,165,940
$12.50
89.0
12.3m
9
Daily Testify Uber Super Expert
3,360,043
$10.08
1
3,360,043
$10.08
54.0
11.2m
10
Copilot CLI Deep Research Agent
3,224,451
$9.67
1
3,224,451
$9.67
42.0
10.0m
Top 10 workflows account for 49.1M tokens (54.3%) of total consumption.
The top 3 single-run workflows each consumed 5.5M+ tokens ($16+ each):
Daily Syntax Error Quality Check β 5.6M tokens in a single run
π Observation: Large codebase scan with 1 turn suggests heavy context loading (probably scanning all Go files). At 18.7 minutes, this is near the timeout limit.
β Recommendation: Consider incremental scanning (changed files only) or filtering to high-risk paths to reduce context size by potentially 50-70%.
Daily Community Attribution Updater β 5.6M tokens, 116 turns
π Observation: The extremely high turn count (116!) indicates a multi-iteration loop processing many contributors. This is the highest-turn workflow in today's run.
β Recommendation: Batch contributor processing rather than individual turns. Consider summarizing contributor data before passing to the agent.
jsweep - JavaScript Unbloater β 5.5M tokens in 22.8 minutes
π Observation: Large JS codebase analysis with 1 turn. High per-run cost suggests full-repo scan.
β Recommendation: Scope to recently modified files or run only when JS files change.
Optimization Opportunities (Detailed)
High Turn Count Workflows
Workflows with many turns consume tokens for each agent iteration:
Workflow
Avg Turns
Potential Savings
Daily Community Attribution Updater
116
Batch processing β 80-90% reduction
Smoke Multi PR
89
Parallel smoke tests β consolidate runs
Dead Code Removal Agent
56
Incremental scanning β 40-60% reduction
Documentation Noob Tester
51
Selective doc testing β 30-50% reduction
Daily Testify Uber Super Expert
54
Targeted file scanning β 30-40% reduction
High-Frequency Workflows (Accumulate Cost Over Time)
These run many times and add up:
Workflow
Runs
Total Tokens
Avg/Run
Action
Issue Monster
18
4.5M
248K
β Already efficient per-run
Auto-Triage Issues
16
5.4M
338K
Consider rate-limiting or batching
Agent Container Smoke Test
9
1.4M
158K
β Efficient per-run
Contribution Check
6
5.4M
902K
Largest frequent consumer β review scope
Historical Comparison
Date
Total Tokens
Runs
Est. Cost
Tokens/Run
2026-01-15
94,899,840
178
~$285
533K
2026-01-22
101,969,175
239
~$306
427K
2026-01-28
200,676,032
281
~$602
714K
2026-02-11
237,834,034
378
~$713
629K
2026-02-19
119,326,605
83
~$358
1,437K
2026-02-20
166,507,574
195
~$500
854K
2026-04-02
90,369,806
101
~$271
895K
Trend: Token volume is significantly down from the February peak. The Feb-11 peak of 237M tokens represented the highest consumption. Today's 90M is the second-lowest recorded, suggesting either fewer workflows running or more efficient runs. The avg tokens/run of 895K remains relatively high.
Methodology & Data Quality Notes
Methodology
Data Source: Pre-downloaded GitHub Actions workflow run audit logs (/tmp/gh-aw/copilot-logs.json)
Engine Filter: Copilot engine workflows only
Runs Analyzed: 101 runs with token data (34 runs had no token data β in-progress or not tracked)
Cost Model: $3.00 per 1M tokens (approximate claude-sonnet average; actual costs vary by model mix and cache usage)
reacted with thumbs up emoji reacted with thumbs down emoji reacted with laugh emoji reacted with hooray emoji reacted with confused emoji reacted with heart emoji reacted with rocket emoji reacted with eyes emoji
Uh oh!
There was an error while loading. Please reload this page.
-
Executive Summary
Over the last 30 days of collected data, Copilot-powered agentic workflows consumed 90,369,806 tokens (~$271.11 estimated*) across 101 workflow runs spanning 50 unique workflows. Today's snapshot covers all completed Copilot runs visible in the audit log.
Key Highlights
π Token Usage & Cost Trends
Historical Token Consumption
Token usage shows a downward trend from the February peaks (~237M tokens on 2026-02-11) to the current 90M. Today's run count of 101 is the second lowest on record, down from a peak of 378 runs.
Estimated Daily Cost Trends
π Top 10 Workflows by Token Consumption
Top 10 workflows account for 49.1M tokens (54.3%) of total consumption.
Per-Workflow Detailed Statistics (All 50 Workflows)
π‘ Insights & Recommendations
High-Cost Workflows
The top 3 single-run workflows each consumed 5.5M+ tokens ($16+ each):
Daily Syntax Error Quality Check β 5.6M tokens in a single run
Daily Community Attribution Updater β 5.6M tokens, 116 turns
jsweep - JavaScript Unbloater β 5.5M tokens in 22.8 minutes
Optimization Opportunities (Detailed)
High Turn Count Workflows
Workflows with many turns consume tokens for each agent iteration:
High-Frequency Workflows (Accumulate Cost Over Time)
These run many times and add up:
Historical Comparison
Trend: Token volume is significantly down from the February peak. The Feb-11 peak of 237M tokens represented the highest consumption. Today's 90M is the second-lowest recorded, suggesting either fewer workflows running or more efficient runs. The avg tokens/run of 895K remains relatively high.
Methodology & Data Quality Notes
Methodology
/tmp/gh-aw/copilot-logs.json)/tmp/gh-aw/repo-memory/default/memory/token-metrics/history.jsonlData Quality Notes
total_cost: 0β costs are estimated from token countsDaily Copilot Token Consumption Reportrun itself (in-progress) is excluded from statsReferences: Β§23897793035
Beta Was this translation helpful? Give feedback.
All reactions