Ordered by predicted speed. Bandwidth sets decode rate; memory decides whether it runs at all.
Every figure here is computed, not measured. Fit is weights at each quantisation against device memory, with a 25% allowance for the KV cache, activations and the OS. Decode rate is the memory-bandwidth roofline at 70% efficiency. Nobody has run this model on these devices.
Device
Memory
Bandwidth
Best quality
Weights
tok/s
Fastest
NVIDIA GB200 Grace Blackwell Superchip
372.0 GB
16000.0 GB/s
bf16
222.42 GB
~50.4
~201.4 q4
AMD Instinct MI355X
288.0 GB
8000.0 GB/s
fp8
111.21 GB
~50.4
~100.7 q4
NVIDIA B200
180.0 GB
7700.0 GB/s
fp8
111.21 GB
~48.5
~96.9 q4
Google TPU7x (Ironwood)
192.0 GB
7380.0 GB/s
fp8
111.21 GB
~46.5
~92.9 q4
AMD Instinct MI325X
256.0 GB
6000.0 GB/s
fp8
111.21 GB
~37.8
~75.5 q4
AMD Instinct MI300X
192.0 GB
5300.0 GB/s
fp8
111.21 GB
~33.4
~66.7 q4
NVIDIA H200 SXM
141.0 GB
4800.0 GB/s
q6
83.41 GB
~40.3
~60.4 q4
NVIDIA H100 NVL
94.0 GB
3900.0 GB/s
q5
69.51 GB
~39.3
~49.1 q4
NVIDIA H100 SXM
80.0 GB
3350.0 GB/s
q4
55.6 GB
~42.2
~42.2 q4
AMD Instinct MI250X
128.0 GB
3200.0 GB/s
q6
83.41 GB
~26.9
~40.3 q4
Google TPU v5p
95.0 GB
2765.0 GB/s
q5
69.51 GB
~27.8
~34.8 q4
NVIDIA A100 80GB SXM
80.0 GB
2039.0 GB/s
q4
55.6 GB
~25.7
~25.7 q4
NVIDIA H100 PCIe
80.0 GB
2000.0 GB/s
q4
55.6 GB
~25.2
~25.2 q4
Apple M3 Ultra
512.0 GB
819.0 GB/s
bf16
222.42 GB
~2.6
~10.3 q4
Apple M2 Ultra
192.0 GB
800.0 GB/s
fp8
111.21 GB
~5.0
~10.1 q4
Apple M4 Max
128.0 GB
546.0 GB/s
q6
83.41 GB
~4.6
~6.9 q4
Apple M2 Max
96.0 GB
400.0 GB/s
q5
69.51 GB
~4.0
~5.0 q4
Apple M3 Max
128.0 GB
400.0 GB/s
q6
83.41 GB
~3.4
~5.0 q4
Where it runs
Taken from the card's availability section.
Platform
Id on that platform
Hugging Face
Qwen/Qwen1.5-110B-Chat
Verified benchmark evidence
Each row was checked against its source by a reviewer, and carries the model identifier as actually evaluated — which is not always the same as this card's.