I moved to MacBook Pro M5 from Apple's M3 Pro model: Here is what changed

Apple's baseline M5 silicon brings clear performance gains to the MacBook Pro, but long-standing limitations such as lack of Face ID and a dedicated numerical layout on keyboard remain unchanged

Don't want to miss the best from Business Standard?

My primary work machine had been the MacBook Pro powered by the M3 Pro chip with 36GB unified memory and 512GB storage. I had used this configuration long enough for it to form a reference for evaluating professional laptops across operating systems. It remained steady during extended workloads, multitasking, browser-led research, media editing, and AI-assisted tasks that were increasingly part of my workflow. Through this experience, I developed a clear sense of the strengths of Apple Silicon and the areas where it imposed constraints.

 

The device had also acted as a benchmark against several Windows 11 machines I tested this year, including those using Arm64 chips from Qualcomm. These comparisons highlighted clear behavioural patterns, such as consistent battery endurance and stable thermal management on the MacBook Pro side, and the varying levels of optimisation across Windows machines. This helped establish a baseline understanding of what a professional-grade laptop should deliver during sustained use.

 

The familiarity with the M3 Pro model also exposed the limitations that had persisted through multiple MacBook generations, including the absence of Face ID, the lack of a numerical keypad, and the memory ceilings that placed constraints on certain high-load operations. These observations formed the reference frame through which the next-generation device would be evaluated.

 

This context was critical when I moved to the new MacBook Pro M5.

Why the M5 was worth resting early

Apple introduced the baseline M5 chip ahead of the Pro and Ultra variants expected next year. The positioning of this first release was significant. I sensed that the model is capable of operating closer to the performance envelope of the previous Pro chips, including the M3 Pro I had been using. This raised a relevant question: could the new baseline reduce the need for stepping up to the Pro tier?

 

  The underlying architectural changes made the chip worth evaluating. Apple increased memory bandwidth from 120GB/s to 153GB/s, added a neural accelerator to every GPU core, and updated the storage system for higher throughput. These changes were targeted not only at general compute tasks but also at the growing category of on-device AI workloads. This was directly relevant to my daily use, which now includes local inference, image generation, data consolidation across spreadsheets, and media-related tasks.

How the M5 was tested

The model used for testing was the entry-level 14-inch MacBook Pro M5 in Silver, configured with 16GB unified memory and 512GB storage. All M5 variants share a uniform architecture: a 10-core CPU, a 10-core GPU, and a 16-core Neural Engine. This consistency eliminated uncertainty in performance expectations and allowed direct comparison with my previous M3 Pro workflow.

 

To assess the behaviour of the new chip, I replicated my complete routine. This included:

• Running large language models in LM Studio
• Generating images in DiffusionBee
• Managing documents, communication tools, and browser-intensive research
• Editing and encoding media files using free-tier Davinci Resolve
• Working with multiple spreadsheets for cross-referencing and data interpretation

To test the limits of the 16GB configuration, I loaded the 20B variant of OpenAI’s open-source model in LM Studio. Tasks were monitored for memory consumption, inference errors, and fallback behaviour. Parallel operations such as video encoding and spreadsheet processing were executed to simulate the pressure conditions under which the M3 Pro had earlier been tested.

 

This provided a like-for-like scenario for measuring differences in performance, stability, and system behaviour.

Where the M5 delivered clear gains

The architectural changes in the M5 chip produced noticeable differences across multiple tasks. The increased memory bandwidth and the integration of neural accelerators on every GPU core changed how local AI workloads responded. Text-based inference ran with lower latency, and image generation completed with greater efficiency. The Neural Engine continued handling everyday on-device tasks within Apple’s Intelligence framework, but the GPU accelerators managed most of the heavier operations.

 

  The improvements were most visible during AI-focused sessions. Large language models responded more quickly and maintained consistency across iterations. Diffusion-based imaging tools showed shorter processing cycles. Media-related tasks that leaned on GPU throughput also benefitted from the increased compute capacity.

 

However, the inherent limits of the 16GB unified memory were unchanged. Running the 20B model in LM Studio continued to trigger out-of-memory errors unless specific fallback features were disabled. Similar constraints appeared when I attempted to run spreadsheet-heavy operations while encoding video files. These scenarios demonstrated that memory ceilings remained a limiting factor regardless of the silicon generation.

 

The M5 also demonstrated gains that could benefit gaming, although this was not a core part of testing. The increase in GPU compute capacity supported Apple’s claims of improved graphics-related performance.

Where the M5 leaves the same gaps

Despite the gains in silicon performance, certain long-standing aspects of the MacBook Pro remained unchanged. The device still does not offer Face ID. Authentication continues to depend on Touch ID, and the absence of presence-based features still limits contextual interactions that could have enhanced workflow convenience.

 

The keyboard layout also remains unchanged, including the absence of a dedicated numerical keypad. This has implications for users working with finance, data entry, or anything involving sustained numerical input. It does not affect all workflows, but for those who require frequent numeric operations, the limitation persists.

 

These gaps had been present in earlier generations and remain unaddressed in the M5 iteration.

Verdict

The MacBook Pro M5 introduces clear improvements in performance, particularly for users who work with on-device AI tools or GPU-driven workloads. These gains make the baseline model suitable for professionals who rely on consistent throughput and efficient processing.

 

However, those who require higher memory capacity or who expect any shift in input design or authentication systems may find that the structural limitations remain. The device delivers improved processing power but does not change the broader experience established by earlier models.

 

  For users prioritising performance improvements, the M5 is a meaningful upgrade. For users waiting for design or feature changes, the next generation may offer more definition.

Apple MacBook Pro M5: Pricing

14-inch MacBook Pro M5 with standard display

• 16GB Unified memory + 512GB SSD storage: Rs 169,900

Apple MacBook Pro M5: Unboxing