Alibaba
Qwen2.5 Coder 7B
code
7,6B
Parameters
4,4 Go
Q4_K_M file
≥ 5,9 Go
Recommended VRAM
32 768
Max context
Speed by graphics card
Simulate in the “Configurator” →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_M | VRAM | Generation | Prompt | VRAM used | Power | € elec. / M tok | Price |
|---|---|---|---|---|---|---|---|
| RTX 5070 Ti | 16 Go | 162 tok/s | 7 492 | 5,9 Go | 138 W | 0,06 €/M tok | — |
| RTX 4070 Ti | 12 Go | 96,7 tok/s | 6 574 | 5,1 Go | 151 W | 0,11 €/M tok | — |
Not measured on these cards yet — click for an estimate:
Technical sheet
- Published
- 2024-09-17
- Architecture
- 28 layers · 4 KV heads · dimension 128
- Attention
- Standard: every layer keeps the whole context in memory.
- Context cache (f16)
- ≈ 55 MB per 1,000 tokens · 1,8 Go for 32,768 tokens
- GGUF file
- bartowski/Qwen2.5-Coder-7B-Instruct-GGUF · Qwen2.5-Coder-7B-Instruct-Q4_K_M.gguf Download (4,4 Go)
- Official repo
- Qwen/Qwen2.5-Coder-7B-Instruct
Public Hugging Face data (official config.json, GGUF repo). The context cache only counts full-attention layers.