Apple MacBook Pro M6 Tests: Geekbench GPU 64,115 vs 48,840 on M5

Voltgrid found no earlier independent review of Apple's 14-inch MacBook Pro with M6. The first two labs to test Apple's 14-inch MacBook Pro with M6 found larger graphics than CPU gains over M5 in Geekbench 6; Engadget's GPU score rose to 64,115 from 48,840.

These are two labs, not three. Engadget published its review on Nov. 5, and Tom's Hardware and Tom's Guide print identical Geekbench and battery results, which points to shared testing at their common publisher, Future; neither says so. Each tested the $1,999 base configuration: a 12-core CPU, a 12-core GPU, 16GB of unified memory and a 1TB SSD. The laptop ships Nov. 6, as Voltgrid reported. Apple claims up to 30% faster graphics than M5, and says up to 25% faster AI workloads such as LLM prompt processing; Engadget paraphrases that as twice as fast. Voltgrid found no lab AI test on the laptop as of Nov. 5.

What the Two Labs Measured in M6 Benchmarks

The labs disagree on the size of the CPU gain. The Tom's lab measured Geekbench 6 at 4,752 single-core and 21,184 multi-core, against 4,288 and 17,926 on its M5 unit, a gain of about 10% and 18%. Engadget's run on the same configuration gave 4,458 and 19,077, against 4,310 and 18,003 on its own M5, which Engadget calls ever-so-slightly faster. Engadget measured a larger step in Cinebench R24, where its M6 was around 30% faster than M5 in multi-threaded work. Neither outlet explains the Geekbench gap, and the results should not be averaged.

On graphics, the Tom's lab measured 93 fps in 3DMark Wild Life Extreme Unlimited, against 73 fps on M5. Engadget saw 90 to 100 fps in Lies of P at 3,024 by 1,890 with maximum settings, about 25 fps more than M5. In a HandBrake transcode from 4K to 1080p, the Tom's lab timed 2:50 against 3:31 on M5.

SSD Speed, Thermals and Battery Life Results

Some things did not move. Tom's Hardware measured a 25GB file copy at 1,914.23 MBps, against 1,917.35 MBps on M5. Over ten back-to-back Cinebench 2026 runs, the score opened at 5,026.55 and held around 5,440, with the chip peaking at 80°C. The lab says its temperature tool does not yet fully support M6.

The battery results come from different tests and cannot be compared. The Tom's lab measured 19 hours 5 minutes of web use at 150 nits, against 18 hours on M5; Tom's Guide lists Apple's rating as 16 hours of wireless web, while Apple's own release says up to 24 hours. Engadget recorded 9 hours 35 minutes in a script that simulates programming work.

Why Memory Bandwidth Sets Local AI Speed

A language model writes its answer one token at a time, and each token needs the model's weights read from memory again. On a laptop, that makes memory bandwidth, not CPU score, the usual limit on how fast a local model replies, while memory capacity decides which models fit at all. Apple lists M6 at up to 32GB and up to 170GB/s, against 307GB/s for M5 Pro.

Apple's AI claim for this laptop is narrower than it sounds. Apple says up to 25% faster AI workloads such as LLM prompt processing, which is the step that reads the prompt, and it publishes no reply-speed figure. Until a lab runs a local model on the M6 MacBook Pro, buyers have the bandwidth figures and Apple's prompt-processing claim, but no measured answer to how fast it writes.

What the M6 MacBook Pro Means for Buyers

The price is the shared complaint: both Engadget and Tom's Guide note the base model rose $400 from $1,599. For anyone running language models locally, memory matters more than CPU scores. Apple lists M6 at up to 32GB and up to 170GB/s, while the M5 Pro, which Engadget prices at $2,499, offers up to 64GB at 307GB/s, a gap set out in Voltgrid's M6 spec file. Voltgrid found no lab local-model test on the laptop as of Nov. 5, so Apple's 25% AI figure remains Apple's claim.