Video summary

MacBook Pro M5 Max vs M4 Max Benchmark Results Are INSANE!

Main summary

Key takeaways

Product Review

Product Reviewed / Compared

  • Apple MacBook Pro M5 Max (new) vs MacBook Pro M4 Max (2024)
    • Both units were sent to the reviewer.
    • Tested in “brand new” condition, with batteries starting at ~100%.

Key Specs Differences Called Out

CPU Cores

  • M4 Max: 16-core (12 performance, 4 efficiency)
  • M5 Max: 18-core (12 performance + 6 “super cores”)

GPU

  • Both: 40-core GPU
  • M5 Max adds: AI-enhanced GPU cores (as described by the reviewer)

Memory

  • Both: 128GB
  • Memory bandwidth: 546 GB/s (M4 Max)614 GB/s (M5 Max)

Storage

  • Both: 4TB SSD
  • Main change attributed to PCIe 4 vs PCIe 3 on the storage path

Wireless

  • M4 Max: Wi‑Fi 6E, Bluetooth 5.3
  • M5 Max: Wi‑Fi 7, Bluetooth 6

Testing Setup / Conditions Mentioned

  • Benchmarks were run after installing software/tests.
  • Batteries set to 100% at the start, with drain measured at the end.
  • Performance tests included:
    • Blackmagic
    • Speedometer 3.1
    • Geekbench
    • Cinebench
    • Geekbench Metal
    • 3DMark / nomad light
    • Real workload tests: Cyberpunk 2077, Blender export, raw→JPEG, ProRes, HEVC
    • AI benchmarks: Geekbench AI, Fusion B2, plus a Llama model test
  • Thermal testing:
    • Cinebench multi-core
    • Checked via Hot app sensors + a thermal camera

Performance Results (M5 Max Improvements)

1) Storage / I/O (Blackmagic + practical browsing proxy)

  • Blackmagic read speed: ~5200 MB/s (M4 Max) → ~13,000 (M5 Max)
    • Claimed improvement: ~150% (about 2.5× faster)
  • Blackmagic write speed: ~7600 (M4 Max) → ~12,800 (M5 Max)
    • Claimed improvement: ~68%
  • Speedometer 3.1: 49 → 60.76
    • Claimed improvement: ~24%
  • Reason given: PCIe 4 on storage (instead of PCIe 3 previously)

2) CPU performance (Geekbench + Cinebench)

  • Geekbench single-core: 4010 → 4278 (~6.7% increase)
  • Geekbench multi-core: 26078 → 29311
    • Reviewer described this as the fastest multi-core Mac result
  • Cinebench single-core: 675 → 732 (~8.4%)
  • Cinebench multi-core: 7829 → 8445 (~~8% faster)
  • Takeaway: gains are real but not massive because the M4 Max was already extremely strong.

3) GPU / graphics performance (Geekbench Metal, game tests)

  • Geekbench Metal: 188,600 → 228,540 (~21.2% increase)
  • Still / nomad light: 109 FPS → 132 FPS
  • Cyberpunk 2077 @ 1460p Ultra
    • Claimed: ~59 FPS (M4 Max)~85 FPS (M5 Max)
    • Improvement: ~44%
  • Reviewer comparison: M5 Max GPU performance is described as “very close to M3 Ultra” performance, despite M3 Ultra having 80 GPU cores vs M5 Max’s 40.

4) On-device AI performance

  • Geekbench AI
    • CPU: ~18% increase
    • GPU: ~13.8% increase
    • “Neural/engine” called out for MPU: still about ~20% overall (despite a claim of same 16-core amount in Apple’s setup)
  • Fusion B2 (image generation)
    • M5 Max: 4.7s
    • M4 Max: 6.3s
  • Llama tokens/sec (same model, same prompt length)
    • M4 Max: ~90.45 tokens/sec
    • M5 Max: ~107 tokens/sec

5) Real production-ish tasks (media/export)

  • Blender export: 25.69s → ~16s
    • Improvement: ~37.6% faster
  • RAW → JPEG: 419s → 383s
    • Improvement: ~8.6%
  • ProRes 422 export: 28.5s → 26s
    • Improvement: ~8.8%
  • ProRes → HEVC export: 128s → 105s
    • Described as “super impressive” (implying a meaningful uplift)
  • Reviewer expectation: future Apple chips (they mention M6) should further improve media engines, ideally with more AI acceleration.

Thermals / Efficiency Results

Temperature under sustained load

  • Average temperature (sensor / Hot app):
    • M4 Max: ~113°C
    • M5 Max: ~105°C
    • Difference: ~8°C cooler
  • Thermal camera results (surface/keyboard + fan exhaust area):
    • Fan exhaust area:
      • M4 Max: ~48.7°C
      • M5 Max: ~46°C
    • Difference: ~2.7°C

Takeaway: more performance while running slightly cooler.

Battery life under their test run

  • Battery remaining after tests:
    • M4 Max: 27% remaining
    • M5 Max: 34% remaining
    • Difference: +7% battery
  • Conclusion: M5 Max is more power efficient under sustained load, even though it’s faster.

Packaging / User Experience Notes

  • Unboxing described as “very normal” (no meaningful differences).
  • Charger included in the box in the UK
    • Reviewer notes US models typically already include it, and contrasts with MacBook Air packaging based on their prior video.

Pros Mentioned (Unique Positive Points)

  • Huge storage gains (Blackmagic read/write, and browsing proxy via Speedometer)
  • Strong CPU improvements (single + multi-core; multi-core described as record/very high)
  • Noticeable GPU uplift in benchmarks and real gaming (~44% faster in Cyberpunk 2077 at 1460p Ultra)
  • Better on-device AI performance (Geekbench AI, Fusion B2 speedup, Llama tokens/sec improvement)
  • Faster creative/media exports (especially Blender)
  • Better thermals (~8°C cooler average) despite higher performance
  • Better sustained battery behavior (+7% remaining)
  • Upgraded connectivity: Wi‑Fi 7 + Bluetooth 6
  • UK packaging includes a charger

Cons Mentioned / Limitations (Unique Negative or Smaller Areas)

  • No major downside stated
    • Relative to M4 Max:
      • CPU gains are described as not huge because M4 Max was already extremely powerful
    • No explicit throttling complaints; only “toasty” as a general observation
  • Media engine improvement is framed as something Apple should update in future generations (they mention M6), implying current-gen constraints.

Overall Verdict / Recommendation (Based on the Video)

  • The reviewer presents the MacBook Pro M5 Max as a substantial upgrade over the M4 Max, especially for:
    • Storage I/O
    • GPU-heavy workloads and gaming
    • On-device AI
    • Sustained efficiency (cooler temps + more battery remaining)
  • Implied recommendation: If you can afford it, the M5 Max is the better choice over the M4 Max—though CPU-only buyers may see smaller relative gains.

Unique Points by Speaker

  • Single speaker (the same reviewer) throughout; no distinct contributors’ viewpoints appear in the subtitles.

Original video