I am experiencing severe and repeatable battery drain on an iPhone 16 Pro running iOS 27.2 beta 2 (24B5089g).
The issue has persisted across several days and multiple sysdiagnose captures. The phone is also noticeably warm during some of these periods.
Battery health is currently 89%, with approximately 622 cycles. I have not found any BatteryPanicShutdown events or an obvious battery hardware failure in the sysdiagnose data.
On October 2, I recorded the following:
- 99% battery at 09:00
- 84% at 09:59
- 60% at 10:59
- 48% at 12:00
- 29% at 12:59
- 22% at 13:23
That is a drop of 77 percentage points in 4h23, with approximately 3h10 of screen-on time.
Maximum battery temperature during that period was approximately 40.4°C.
PowerLog also shows repeated high instantaneous power samples:
- 26 samples above approximately 10 W
- maximum sampled value around 21 W
I understand that these are instantaneous samples and not integrated energy measurements.
One particularly interesting period happened while the display was suppressed:
- approximately 10:59–11:08
- battery dropped from 60% to 58%
- calculated instantaneous power median around 2.1 W
- maximum around 2.65 W
- approximately 2 GB of outbound Wi-Fi traffic attributed to
com.apple.nsurlsessiondbetween 11:02:54 and 11:07:55 - only 3 TGI requests occurred during that period, with a combined duration of approximately 1.7 seconds
Across the full 09:00–13:23 interval, PowerLog attributes approximately 8.6 GB of outbound Wi-Fi traffic to nsurlsessiond.
Photos also shows significant network activity separately, but I cannot establish that all nsurlsessiond traffic belongs to Photos or iCloud.
There is also persistent local inference activity.
The ModelCatalog dump reports:
IFP Enabled: false
During the same morning period I counted approximately:
- 349 TGI requests
- ~4m20 combined TGI request duration
- 211 ANE model load events
- 207 ANE model unload events for
com.apple.tgondeviceinferenceproviderservice_platform - 179 load / 170 unload events for
mediaanalysisd
However, I do not think TGI alone explains the battery drain. On October 1, the TGI request rate was very similar (~83 requests/hour vs ~79 requests/hour on October 2), while the battery drain rate was different.
Another important observation is that the phone can reach very low idle power consumption.
Between approximately 05:00 and 09:00, with the display suppressed and after excluding the first 60 seconds following display transitions, the median calculated instantaneous power was around 0.06 W.
This suggests the phone is capable of entering a low-power idle state, and the excessive drain appears to be episodic rather than permanent.
For comparison, on October 1:
- 71% → 15% between 11:18 and 15:02
- approximately 56 percentage points lost in 3h44
- approximately 2h35 screen-on time
- maximum battery temperature around 38°C
- 309 TGI requests during the interval
The workload was different on both days, so I am not treating this as a controlled comparison.
I have several complete sysdiagnose archives available from the affected periods and can provide specific PowerLog queries or additional logs if useful.
I am currently planning to restore the phone to the latest public iOS release and configure it as a new iPhone to determine whether the behaviour is specific to this beta installation/state.
Has anyone seen similar behaviour on iPhone 16 Pro or iOS 27 / 27.2 beta, particularly involving:
- unusually high background
nsurlsessiondtraffic - repeated ANE/TGI model loading and unloading
- severe battery drain despite normal low-power idle behaviour
- recurring heating without a BatteryPanicShutdown or obvious battery hardware fault
For help with battery questions, please post in the Apple Support Community. The Apple Developer Forums are for questions about APIs and features intended specifically for developers.