卖什么
Combining open-source AI models for better answers at lower cost
商业模式拆解
买家：谁掌握预算？
未核实ML or platform engineering lead paying large LLM API bills.
工作流：什么输入变成什么输出？
Trains a lightweight surrogate from the LLM production traces; an acceptor serves predictable inputs locally and defers uncertain cases to the LLM, exposed via an OpenAI-compatible endpoint; also Echo, an ensemble of specialist open models behind one endpoint.
经济账：省了什么，怎么收费？
Cost savings value claim: 95% saved vs GPT-5 at Obside with 99.9% routed accuracy; ~50% agent cost drop at getclaw; open-source TRACER OSS (MIT) and paid Echo.
渠道：怎么接触到客户？
Open source pip install tracer-llm plus Book a call.
融资历史与报道
| 融资日期 | 轮次 | 金额 | 投资方 | 来源 |
|---|---|---|---|---|
| — | YC 2026 夏季 | — | Y Combinator | ycombinator.com |
档案汇总自公开报道和公司官网。商业模式部分是 MaxBC 对这些材料的解读，不是已确认的事实；未知之处留空，不作猜测。 更新于 2026-10-12.