Qwen3-30B-A3B is a smaller Mixture-of-Experts (MoE) model from the Qwen3 series by Alibaba, with 30.5 billion total parameters and 3.3 billion activated parameters. Features hybrid thinking/non-thinking modes, support for 119 languages, and enhanced agent capabilities. It aims to outperform previous models like QwQ-32B while using significantly fewer activated parameters.
发布日期2025年4月29日
参数规模30.5B
上下文长度128K
许可证Apache 2.0
知识截止—
Benchmarks
评测成绩
| 评测基准 | 类别 | 分数 | 来源 |
|---|---|---|---|
| Arena Hard | general reasoning creativity | 91.0 | 来源 |
| AIME 2024 | math reasoning | 80.4 | 来源 |
| LiveBench | math reasoning general | 74.3 | 来源 |
| Multi-IF | reasoning communication language | 72.2 | 来源 |
| AIME 2025 | math reasoning | 70.9 | 来源 |
| BFCL | general reasoning | 69.1 | 来源 |
| GPQA | reasoning general | 65.8 | 来源 |
| LiveCodeBench | reasoning general code | 62.6 | 来源 |
Pricing
API 价格对比
| 服务商 | 输入价 | 输出价 | 上下文 | 吞吐(tok/s) | 延迟(s) | 函数调用 | 代码执行 | 联网搜索 |
|---|---|---|---|---|---|---|---|---|
| Helicone | $0.08 | $0.29 | 41K | — | — | ✓ | ✗ | ✗ |
| 302.AI | $0.11 | $1.08 | 128K | — | — | ✓ | ✗ | ✗ |
价格单位：美元/百万 token，数据来自社区整理，仅供参考。