Qwen2.5 7B Instruct
Qwen · released 2024-10-16 · apache-2.0 license
Qwen2.5-7B-Instruct is the popular small Qwen2.5 model (Apache-2.0): vendor GPQA 36.4, MMLU-Pro 56.3, LiveCodeBench 28.7, outperforming Gemma2-9B and Llama3.1-8B on most vendor benchmarks.
Key specs
| Type | Local open-weight |
|---|---|
| Parameters | 7.62B total |
| Architecture | qwen2 |
| Context window | 131K tokens |
| Knowledge cutoff | 2024-06-30 |
| Modalities | text |
| Recommended backends | — |
| Minimum viable rig | — |
Benchmark scores
| GPQA Diamond | 36.4% |
|---|---|
| SWE-bench Verified | — |
| AIME | — |
| MMLU-Pro | 56.3% |
| BFCL v3 (tool use) | — |
| Composite score | 5.36 |
| Community rating | No reviews yet |
VRAM & disk per quantization
| Quant | VRAM | Disk | RAM | Context |
|---|---|---|---|---|
| Q4_K_M | 5.9 GB | 4.4 GB | — | 131K |
Strengths & weaknesses
Strengths: Strong small-model math (vendor MATH 75.5); Beats Gemma2-9B and Llama3.1-8B on most vendor evals; Apache-2.0
Weaknesses: IFEval (71.2) below Llama3.1-8B; No native reasoning mode