# Condense.chat 开放压缩代理，降低AI编程智能体token成本

- 来源：🚨 AI News | TestingCatalog (@testingcatalog)
- 发布时间：2026-07-03 21:09
- AIHOT 分数：69
- AIHOT 链接：https://aihot.virxact.com/items/cmr4z9nu806dwsll5zijz6van
- 原文链接：https://x.com/testingcatalog/status/2073031325246726196

## AI 摘要

Condense.chat 推出位于 AI 编程智能体与模型之间的压缩代理，可降低 token 费用。其内置两款自研模型：Helene 1 在缓存前剔除不必要 token；Adeline 1 将智能体循环压缩至原大小约 9%。无需修改系统提示词。引用推文展示真实会话基准测试：Condense 节省 53.8% 的 token，账单降低 37.3%；竞品 headroom 仅为 15.1% 和 13.8%。官方提供 1 亿 free token 试用。

## 正文

Condense[.]chat has opened access to its new compression proxy, which sits between a coding AI agent and the model, cutting the token bill.

Two in-house models handle the work:
Helene 1 strips unnecessary tokens before they hit cache.
Adeline 1 compacts settled agent loops down to about 9% of their size.

Condense requires no change to the system prompt, so the tool's results remain untouched.

### 引用推文

> condense.chat：We're giving you 100M free tokens to prove your agent is wasting context. Most of what your agent sends upstream is dead weight. We built state-of-the-art compa...