Alibaba

Qwen3.5 2B

2,3B

Parameters

1,3 Go

Q4_K_M file

≥ 2,8 Go

Recommended VRAM

262 144

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 Go394 tok/s18 8842,5 Go107 W0,02 €/M tok—

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

Technical sheet

Published
2026-02-28
Architecture
24 layers · 2 KV heads · dimension 256
Attention
Hybrid: 6 of 24 layers use standard attention, the others a fixed-memory mechanism (Mamba, linear attention) — a very light cache.
Context cache (f16)
≈ 12 MB per 1,000 tokens · 0,4 Go for 32,768 tokens
GGUF file
bartowski/Qwen_Qwen3.5-2B-GGUF · Qwen_Qwen3.5-2B-Q4_K_M.gguf Download (1,3 Go)
Official repo
Qwen/Qwen3.5-2B

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