Apple-M5-or-Apple-M4
The M5 is smoking, but we expect it won't get too hot. This month is the month for the M5's launch and release. While performance figures fly around with every new generation processor announcement, the question that needs to be asked is: How much will this power difference be felt in daily use?
The M5 series, especially with the new accelerators added to the graphics unit and higher memory bandwidth features, is remarkable. Despite this massive power increase, the M5 retains the famous coolness of the M4.
M4 Pro and M5 Pro Difference: What Will You Feel as a User?
| Features |
M4 Pro Series |
M5 Pro Series |
User Experience |
| Battery Life |
18-22 Hours |
20-24h (+2) |
Forget your charger; work all day without an outlet. |
| Display |
Liquid Retina XDR |
Tandem OLED |
Crystal clear visuals even in direct sunlight. |
| Connection |
Wi-Fi 6E |
Wi-Fi 7 |
Stable internet that doesn't drop in crowded cafes. |
| Responsiveness |
Very Fast |
15-20% Agile |
Nimbly handles heavy compilation or large files. |
| Multitasking |
120 GB/s |
153 GB/s |
No lagging with Render + Zoom + 50 Tabs. |
| SSD Speed |
2.8 GB/s |
6.7 GB/s (2.4x) |
Copying projects 2.4x faster than M4. |
| AI Ingest (PPS) |
~1,855 PPS |
~4,468 PPS |
4x faster reading and analyzing large prompts and files.
|
Technical Note: What's the Source of This Speed?
The increase in the M5 isn't just a "software improvement." Apple has increased Memory Bandwidth by 30%, widening the path of data to and from the processor. This allows the computer to "breathe," preventing stuttering, especially when playing multiple 8K streams simultaneously or running large AI models natively.
Should you buy it or wait?
- M4 Choice: If you do daily office work and rarely edit videos, the M4 Pro is still the most sensible "price/performance" choice right now.
- M5 Choice: If you are a professional software developer, 3D artist, or a power user who constantly pushes hardware limits. It is a vital investment for those who frequently work with large data files and/or run local LLMs with too large inputs that demand peak bandwidth.