You should probably go give @poolsideai a follow on Hugging Face. These folks are on a roll, releas...
TL;DR · AI 摘要
Poolside AI的Laguna S2.1是当前在本地运行的最佳编码模型之一，其在Terminal-Bench 2.1和DeepSWE基准测试中表现优异。
核心要点
- Laguna S2.1在Terminal-Bench 2.1得分70.2，超越5-25倍参数量的模型
- Poolside AI每月发布一个代理编码模型，Laguna S2.1支持单DGX Spark/Mac本地运行
- DeepSWE基准测试显示其在长周期任务中表现突出
结构提纲
按章节快速跳转。
- §模型推荐
Thomas Wolf推荐关注Poolside AI的Laguna S2.1编码模型。
- ·性能表现
Laguna S2.1在Terminal-Bench 2.1得分70.2，超越同规模模型。
- ›基准测试
在DeepSWE长周期任务中表现优于多个大模型。
- ·发布频率
Poolside AI每月发布一个代理编码模型，持续迭代优化。
思维导图
用一张图看清主题之间的关系。
查看大纲文本（无障碍 / 无 JS 友好）
- Laguna S2.1模型介绍
- 性能优势
- Terminal-Bench 2.1得分70.2
- 超越5-25倍参数量模型
- 技术特性
- 支持本地运行（DGX Spark/Mac）
- DeepSWE基准测试通过
- 开发团队
- Poolside AI每月发布新模型
金句 / Highlights
值得收藏与分享的关键句。
Laguna S2.1在Terminal-Bench 2.1得分70.2，与5-25倍参数量模型并列
支持单DGX Spark或Mac本地运行，无需云端资源
DeepSWE基准测试验证其在长周期任务中的可靠性
Thomas Wolf on X: "You should probably go give @poolsideai a follow on Hugging Face. These folks are on a roll, releasing one agentic coding models every month and the latest one Laguna S2.1 is quite possibly the best coding model you can run locally (single DGX spark or Mac) atm https://t.co/74uQy16ksa" / X
Thomas Wolf
@Thom_Wolf
You should probably go give
@
poolsideai
a follow on Hugging Face. These folks are on a roll, releasing one agentic coding models every month and the latest one Laguna S2.1 is quite possibly the best coding model you can run locally (single DGX spark or Mac) atm
Poolside
@poolsideai
Jul 21
Replying to
Laguna S 2.1 is, as far as we can measure, the most capable agentic coding model in its weight class. On Terminal-Bench 2.1 it scores 70.2, sitting beside models 5–25x its size and ahead of several of them. And on DeepSWE from
datacurve
, the hardest long-horizon benchmark we
Show more
8:46 AM · Jul 22, 2026
11.5K
Views
12
0
1
2
17
7
122
16
6