---

细节

---

---

Squire 主要特点

• Run History
  Squire saves metrics and history of model runs for review.
• Web Dashboard
  You can create and manage workloads using the online dashboard.
• Workload Keys
  You can create a workload and copy a special key to identify it.
• Prompt Privacy
  Your prompts and outputs are stored briefly and then deleted to keep data private.
• Decision Making
  Squire gives you a recommendation to switch models, test more, or stay with the current one.
• Production Path
  Your app runs normally without changes, keeping your path unchanged.
• Decision Viewing
  You can see the decision through the dashboard or by calling an API.
• Model Evaluation
  Squire compares models in the background and suggests whether to switch or not.
• Background Testing
  Squire runs tests while your app is working, without stopping it.
• Shadow Data Sending
  Your app sends requests normally, and Squire observes the outputs after they are processed.
• Metrics Only by Default
  Squire only collects metrics unless you choose to make changes.
• Automatic Queue Clearing
  Raw data waiting to be tested clears automatically after a while.
• Model Switching Decisions
  Squire tests different AI models in the background and tells you when to switch based on safety.
• Simple Python Integration
  You can add a background request in your code so your app still works if Squire fails.

---

徽章

---

Squire 替代品

---

使用的技术

查看所有技术