Qwen 3 8B vs Qwen 2.5 7B
Architecture Comparison
SpecQwen 3 8BQwen 2.5 7B
TypeDENSEDENSE
Total Parameters8.2B7.6B
Active Parameters8.2B7.6B
Layers3628
Hidden Dimension4,0963,584
Attention Heads3228
KV Heads84
Context Length131,072131,072
Precision (default)BF16BF16
Memory Requirements
PrecisionQwen 3 8BQwen 2.5 7B
BF16 Weights16.4 GB15.2 GB
FP8 Weights8.2 GB7.6 GB
INT4 Weights4.1 GB3.8 GB
KV-Cache / Token147456 B57344 B
Activation Estimate1.00 GB1.00 GB
Minimum GPUs Needed (BF16)
H100 SXM1 GPU1 GPU
L40S1 GPU1 GPU
Quality Benchmarks
BenchmarkQwen 3 8BQwen 2.5 7B
Overall6770
MMLU72.074.2
HumanEval42.042.8
GSM8K78.082.0
MT-Bench77.079.0
Qwen 3 8B
MMLU
72.0
HumanEval
42.0
GSM8K
78.0
MT-Bench
77.0
Qwen 2.5 7B
MMLU
74.2
HumanEval
42.8
GSM8K
82.0
MT-Bench
79.0
Capabilities
FeatureQwen 3 8BQwen 2.5 7B
Tool Use✓ Yes✓ Yes
Vision✗ No✗ No
Code✓ Yes✓ Yes
Math✓ Yes✓ Yes
Reasoning✓ Yes✗ No
Multilingual✓ Yes✓ Yes
Structured Output✓ Yes✓ Yes
API Pricing Comparison
Cheapest Output (Qwen 3 8B)
$0.20/M
Input: $0.20/M
Cheapest Output (Qwen 2.5 7B)
$0.20/M
Input: $0.20/M
| Provider | Qwen 3 8B In $/M | Out $/M | Qwen 2.5 7B In $/M | Out $/M |
|---|---|---|---|---|
| together | $0.20 | $0.20 | $0.20 | $0.20 |
| fireworks | $0.20 | $0.20 | $0.20 | $0.20 |
Recommendation Summary
- ‣Qwen 2.5 7B scores higher on overall quality (70 vs 67).
- ‣Qwen 2.5 7B has a smaller memory footprint (15.2 GB vs 16.4 GB BF16), making it easier to deploy on fewer GPUs.
- ‣Qwen 2.5 7B is stronger at code generation (HumanEval: 42.8 vs 42.0).
- ‣Qwen 2.5 7B is better at math reasoning (GSM8K: 82.0 vs 78.0).