Nvidia check failed: Not found (404) - cc-switch
Quick fix
Upgrade cc-switch to v3.16.3+; new version uses lightweight HTTP probing, fixing NVIDIA interface false 404 errors.
Symptom
Section titled “Symptom”Nvidia 检查失败: Not found (404)Nvidia check failed: Not found (404)
In cc-switch v3.16.2 and earlier, the 'Manage & Speed Test' feature sent real model requests or accessed specific API paths (e.g., `/v1/models`) for connectivity testing. Since the NVIDIA API (`integrate.api.nvidia.com`) does not support these test endpoints, it returns a 404 error directly, causing actual inference to be falsely reported as unavailable. Some users may see variant errors like `Nvidia 检查失败: HTTP 404: 404 page not found`, with the same root cause.
Upgrade cc-switch to v3.16.3 or later.
Re-enter the NVIDIA provider edit page and click 'Manage & Speed Test'. In v3.16.3, connectivity checks changed to directly GET the provider's `base_url`; any HTTP response is considered reachable, eliminating false 404 reports.
Affected Versions
Section titled “Affected Versions”Source Issues
Section titled “Source Issues”This page is distilled from 2 real issues
- Actual model calls work fine, why does speed test still report an error?
- This is a speed test mechanism issue in older versions of cc-switch; it requested endpoints unsupported by NVIDIA. As long as actual inference works normally, you can use it normally. Upgrading to v3.16.3 is recommended to eliminate false reports.
- What should I do if 404 errors persist after enabling local routing mapping?
- Enabling routing mapping cannot resolve the issue of old versions requesting wrong endpoints during speed tests. This is a software bug; please upgrade cc-switch directly to v3.16.3.
Related problems
Section titled “Related problems”This is an unofficial community wiki with no affiliation to the cc-switch authors or the project itself. Its content is compiled from the project's public GitHub issues. This site distributes no software.