Total processed
632.10M
Input, cache and output tokens
V4.1 Flash · Direct API simulation
632 million tokens.
Only $6.61.
A request-level replay of every DeepSeek turn in your retained OMP sessions, priced against the official API schedule with real UTC peak windows and observed cache hits.
Cache hit rate
98.64%
Of all billed input tokens
Model output
4.05M
Thinking and answer tokens
30-day pace
≈ $33
At the observed six-day cadence
Context economy
The token ocean
Cached context dominates the workload—and barely moves the bill.
619.53Mcached input tokens
Cached input · 619.53MUncached · 8.53MOutput · 4.05M
Invoice anatomy
Where $6.61 went
Output is the largest cost despite being under 1% of processed tokens.
$6.61total estimate
Uncached input22.1% of spend
$1.459
Cache reads33.1% of spend
$2.190
Output44.8% of spend
$2.960
Observed cache behavior
$6.61
Direct API estimate when DeepSeek reproduces the cache hits reported by the actual OMP traffic.
Zero-cache failure case
$113.94
If every repeated context token were billed as a cache miss. Observed caching avoided approximately $107.33.
Spend velocity
Daily direct API estimate
Request timestamps priced against DeepSeek’s UTC peak schedule.
$3.00$2.00$1.00$0