~/gpu/rtx-4060-ti vs rtx-4080

RTX 4060 Ti 16GB manufacturerRTX 4060 Ti 16GBvsRTX 4080 manufacturerRTX 4080

Head-to-head for local LLM inference. The honest comparison: VRAM, bandwidth, compute, and which of the 30 catalog models actually fit on each.

The specs.

$ diff specs rtx-4060-ti rtx-4080
Stat
rtx-4060-ti
rtx-4080
Δ
VRAM
16 GB
16 GB
0%
Memory bandwidth
288 GB/s
717 GB/s
+149%
FP16 compute
175 TFLOPS
390 TFLOPS
+123%
Weights budget at 8K ctx
12 GB
12 GB
0%

Model fit difference.

$ models that change with the card
Fits on both
16of 30
Only on rtx-4060-ti
0
Only on rtx-4080
0

// showing 12 of 30 models; differing fits first

Model
rtx-4060-ti
rtx-4080
Llama 3.2 1B1.23B
fitsFP16/BF16
fitsFP16/BF16
Llama 3.2 3B3.21B
fitsFP16/BF16
fitsFP16/BF16
Llama 3.1 8B8B
fitsQ8_0
fitsQ8_0
Llama 3.3 70B70.6B
overQ4_K_M
overQ4_K_M
Llama 3.1 405B405B
overQ4_K_M
overQ4_K_M
Qwen 2.5 7B7B
fitsQ8_0
fitsQ8_0
Qwen 2.5 32B32.5B
overQ4_K_M
overQ4_K_M
Qwen 2.5 Coder 32B32.5B
overQ4_K_M
overQ4_K_M
Qwen 2.5 72B72.7B
overQ4_K_M
overQ4_K_M
Qwen3 30B A3B30.5B
overQ4_K_M
overQ4_K_M
Qwen 3.5 9B9B
fitsQ8_0
fitsQ8_0
Qwen 3.6 27B27B
fitsQ3_K_M
fitsQ3_K_M

Which one wins for…

$ ./recommend --by-workload
More VRAM headroom

Tied at 16 GB. Choose on bandwidth or price.

Faster decode (bandwidth)

RTX 4080 by +149%.

Faster prefill (compute)

RTX 4080 by +123% TFLOPS.

Catalog models that fit

Tied: 16 of 30 fit on each.

Drill into either card.

$ ./vrambudget --gpu
$ openRTX 4060 Ti 16GB16GBada$ openRTX 408016GBada

Discussion.

$ gh discussion list

// sign in with github to leave a comment. threads live in the repo's discussions tab.