Alibaba

Qwen2.5 3B

3,1B

Parameters

1,8 Go

Q4_K_M file

≥ 3,3 Go

Recommended VRAM

32 768

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 Go305 tok/s14 7813,3 Go103 W0,02 €/M tok—

Not measured on these cards yet — click for an estimate:

Technical sheet

Published
2024-09-17
Architecture
36 layers · 2 KV heads · dimension 128
Attention
Standard: every layer keeps the whole context in memory.
Context cache (f16)
≈ 35 MB per 1,000 tokens · 1,1 Go for 32,768 tokens
GGUF file
bartowski/Qwen2.5-3B-Instruct-GGUF · Qwen2.5-3B-Instruct-Q4_K_M.gguf Download (1,8 Go)

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