Online Test Detailed Operation Guide
Online testing through lmspeed.net consists of four standardized steps:
- Parameter configuration phase: Fill in the required parameters in the web form, including
baseUrl(e.g., https://api.deepseek.com/v1), a valid API key, and a targetmodelId - Test execution phase: The system automatically performs five rounds of stress tests after clicking the Start button, during which TPoS graphs and latency heat maps are displayed in real time.
- Results analysis phase: automatically generate a visualization board at the end of the test, containing 12 core indicators such as response time distribution graphs, first token latency comparisons, etc.
- Report management phaseSupport for exporting results as PDF reports with test metadata or sharing them with team members via temporary links.
For high-frequency test users, the recommendedURL Parameterized CallsShortcut: Construct a link containing test parameters directly in the browser address bar, and the system will automatically recognize and start the test process. However, it is necessary to pay attention to the security specification, and it is recommended that sensitive API keys be configured through environment variables instead of URL explicit transmission.
This answer comes from the articleLM Speed: Rapidly Test Large Model API PerformanceThe
May not be reproduced without permission:AI productivity tools " How to use LM Speed for online API performance testing?
