Qwen3 14B
Open weightsDeprecatedGeneral LLMReasoningTool use
Qwen3 14B is an open-weights instruction model from Alibaba's Qwen team for multilingual chat, reasoning, and tool use, with a 131,072-token context window and text-in/text-out.
128Kcontext
$0.35 / $1.4per 1M tokens (in / out)
14.8Bparameters
7stories in our feed
Specs
hover a value to see its sourceContext window128K
Max output8K
Parameters14.8B
Knowledge cutoff2025-04
Licenceapache-2.0
Commercial useCommercial OK
InputText
OutputText
HF downloads (30d)9M
Familyqwen
API price
USD per 1M tokensInput$0.35
Output$1.4
Cached input—
No price change recorded since we started tracking.
News timeline
from the AITOP feed, one story per event2026-10-12
Task Context 如何改变语言模型中共享注意力头的角色arXiv cs.AI
2026-10-10
2026-10-06
研究：特定起始 token 提示可让 base 模型接近 RL 训练后的推理表现arXiv cs.AI
2026-09-30
MetaCtrl: Metacognitive Controller for LRMsarXiv: DeepSeek
2026-09-29
arXiv 论文：可再推导性决定多级智能体流水线故障后的恢复能力arXiv cs.LG
2026-08-17
PACE-Bench：动态环境下基于代码演进的物理适应基准arXiv cs.AI
2026-08-05
近似投机解码ASD：免训练加速自回归生成arXiv: DeepSeek