Qwen2.5 72B Instruct vs Qwen3 VL 8B Instruct
Qwen3 VL 8B Instruct is the cheaper option on input by 31%; Qwen2.5 72B Instruct is cheaper on output by 47%. Figures below are per million tokens, from vendor pricing cited in our reporting.
| Qwen2.5 72B Instruct | Qwen3 VL 8B Instruct | |
|---|---|---|
| Input per 1M tokens | $0.36 | $0.25 |
| Output per 1M tokens | $0.40 | $0.75 |
| Cached input per 1M | — | $0.12 |
| Context window | 33K tokens | 262K tokens |
| Parameters | — | — |
| Licence | open weights | open weights |
| Benchmark | — | — |
| Vendor | Alibaba | Alibaba |
| Released | — | — |
What the numbers say
- Input price gap: 31% ($0.36 vs $0.25).
- Blended three-to-one input/output rate: $1.48 for Qwen2.5 72B Instruct, $1.50 for Qwen3 VL 8B Instruct.
- Context: 33K tokens against 262K tokens.
- Licence: Qwen2.5 72B Instruct is open weights; Qwen3 VL 8B Instruct is open weights.
Sources
Qwen2.5 72B Instruct specs · Qwen3 VL 8B Instruct specs · Full price index