Meta

Llama 3.2 3B

small machines

3,2B

Parameters

1,9 Go

Q4_K_M file

≥ 3,4 Go

Recommended VRAM

131 072

Max context

Standard llama-bench run (prompt 512 / generation 128 tokens), model fully in VRAM. Click a card for details. Electricity cost at the default price (0,25 €/kWh).

Card · Q4_K_MVRAMGenerationPromptVRAM usedPower€ elec. / M tokPrice
RTX 5070 Ti16 Go270 tok/s13 9703,7 Go111 W0,03 €/M tok—
RTX 4070 Ti12 Go189 tok/s11 5903,3 Go126 W0,05 €/M tok—

Technical sheet

Published
2024-09-18
Architecture
28 layers · 8 KV heads · dimension 128
Attention
Standard: every layer keeps the whole context in memory.
Context cache (f16)
≈ 109 MB per 1,000 tokens · 3,5 Go for 32,768 tokens
GGUF file
bartowski/Llama-3.2-3B-Instruct-GGUF · Llama-3.2-3B-Instruct-Q4_K_M.gguf Download (1,9 Go)

Public Hugging Face data (official config.json, GGUF repo). The context cache only counts full-attention layers.