Apple M2 Ultra vs Apple M5 Ultra

The M5 delivers 22% more FP32 throughput than the M2 Ultra (33 vs 27 TFLOPS dense). The M5 also carries 320 GB more memory (512 GB vs 192 GB).

M2 Ultra: Apple M2, 2023 M5: Apple M5, 2026
FP32 dense lead
+22%
M5: 33 vs 27 TFLOPS
Memory
192 vs 512 GB
320 GB more for the M5
Bandwidth
800 GB/s vs 1.2 TB/s
M5 moves data faster
Download comparison sheet

Share this page

Apple M2 Ultra vs Apple M5 Ultra

A post with this link shows the card on the Image tab as its preview.

Printable version →

Apple M2 Ultra

Full specs →

What is the Apple M2 Ultra?

The M2 Ultra is an APU from Apple on the Apple M2 architecture, an processor that pairs CPU and GPU on one chip. It is a system-on-chip, so it is built into edge and embedded devices rather than sold as a card. M2 Ultra full specifications →

Can I rent the M2 Ultra in the cloud?

Flopper is not tracking any live rental listings for the M2 Ultra yet. The provider directory lists the clouds that rent comparable accelerators on demand. GPU cloud providers →

How does the M2 Ultra compare with other accelerators?

This page sets the M2 Ultra against the M5. The M2 Ultra is the older generation of the two. Both are Apple parts, and both ship as a system-on-chip, so they fit the same class of system. The compare tool builds the same comparison for any two parts in the catalogue. All GPU comparisons →

Where do the M2 Ultra figures come from?

Apple published the M2 Ultra figures in marketing material rather than a datasheet, and Flopper has no independent confirmation of them. How Flopper verifies specifications →

Apple M5 Ultra

Leads on FP32 Full specs →

What is the Apple M5?

The M5 is an APU from Apple on the Apple M5 architecture, an processor that pairs CPU and GPU on one chip. It is a system-on-chip, so it is built into edge and embedded devices rather than sold as a card. M5 full specifications →

Can I rent the M5 in the cloud?

Flopper is not tracking any live rental listings for the M5 yet. The provider directory lists the clouds that rent comparable accelerators on demand. GPU cloud providers →

How does the M5 compare with other accelerators?

This page sets the M5 against the M2 Ultra. The M5 is the newer generation of the two. Both are Apple parts, and both ship as a system-on-chip, so they fit the same class of system. Flopper also compares it head to head with the M3 Ultra, M5 Max and DGX Spark, across every published precision, memory and power. M5 vs M3 Ultra →

Where do the M5 figures come from?

Apple published the M5 figures in marketing material rather than a datasheet, and Flopper has no independent confirmation of them. How Flopper verifies specifications →

Specifications

M2 Ultra
M5
Architecture
Apple M2Apple M5
Launch Year
20232026
Form Factor
SoCSoC
Memory
192 GB512 GB
Memory Bandwidth
800 GB/s1.2 TB/s
TDP
——
Process Node
5nm3nm

Performance (TFLOPS)

M2 Ultra
M5
No verified data available No verified data available
27 TFLOPS 33 TFLOPS
No verified data available No verified data available
No verified data available No verified data available
No verified data available No verified data available
No verified data available No verified data available
FP6
No verified data available No verified data available
No verified data available No verified data available
No verified data available No verified data available

Figures are dense. Where a vendor publishes a figure it sits under the dense one and never decides a winner. Hover any precision for its definition, or read the GPU glossary.

FLOPS by Precision

What actually differs

The M5 is the newer part: Apple M5 (2026) against the M2 Ultra's Apple M2 (2023).

The M5 carries 320 GB more memory (512 GB against 192 GB), so larger models fit on one GPU before you shard. Bandwidth is 800 GB/s against 1.2 TB/s, which sets the ceiling on decode speed.

Weight the FP8 and FP16 rows for LLM work and the FP64 row for HPC.

New to the terms? The glossary explains FP32, Sparsity and TFLOPS.

Frequently asked questions

What is better, the M2 Ultra or the M5?

The M5, on throughput. At FP32 precision the M5 reaches 33 TFLOPS dense against 27 TFLOPS for the M2 Ultra. The performance table on this page lists every published precision for both GPUs.

Which should you choose, the M2 Ultra or the M5?

Choose the M5 for training and high-throughput inference: it has 22% more FP32 throughput than the M2 Ultra. It also carries 320 GB more memory, so larger models fit on one GPU before you shard.

Which has more memory, the M2 Ultra or the M5?

The M5 carries 512 GB of memory versus 192 GB for the M2 Ultra. Memory bandwidth is 800 GB/s for the M2 Ultra and 1.2 TB/s for the M5.

Get Comparison Updates

New GPUs added weekly. Be the first to see how they compare.

© 2026 Flopper.io - Compare the GPUs Powering AI