背景
注册了 nvidia , win11
build.nvidia.com
Try NVIDIA NIM APIs
Experience the leading models to build enterprise generative AI apps now.
想在claude code中试下 nvidia的模型
做了如下配置
cc-switcher
{
"env": {
"ANTHROPIC_BASE_URL": "http://127.0.0.1:15721",
"ANTHROPIC_AUTH_TOKEN": "PROXY_MANAGED"
}
}
启动Claude code
问你是什么模型
从clash verge中看到 claude.exe 还是使用了 主机 : api.anthropic.com:443
从cc-switcher的log看
>>> 请求 URL: https://integrate.api.nvidia.com/v1/chat/completions (model=qwen/qwen3-coder-480b-a35b-instruct)
[2026-04-28][23:55:00][WARN][cc_switch_lib::proxy::forwarder] [claude] [FWD-003] Provider Nvidia 请求失败: 上游 HTTP 429: {"status":429,"title":"Too Many Requests"}
[2026-04-28][23:55:32][INFO][cc_switch_lib::proxy::forwarder] [Claude] >>> 请求 URL: https://integrate.api.nvidia.com/v1/chat/completions (model=qwen/qwen3-coder-480b-a35b-instruct)
[2026-04-28][23:55:34][WARN][cc_switch_lib::proxy::circuit_breaker] [CB-004] 熔断器触发: 连续失败 8 次 → Open
[2026-04-28][23:55:34][WARN][cc_switch_lib::proxy::forwarder] [claude] [FWD-003] Provider Nvidia 请求失败: 上游 HTTP 503: ResourceExhausted: All workers are busy, please retry later
[2026-04-28][23:56:12][INFO][cc_switch_lib::proxy::forwarder] [Claude] >>> 请求 URL: https://integrate.api.nvidia.com/v1/chat/completions (model=qwen/qwen3-coder-480b-a35b-instruct)
[2026-04-29][00:01:29][INFO][cc_switch_lib::proxy::forwarder] [Claude] >>> 请求 URL: https://integrate.api.nvidia.com/v1/chat/completions (model=qwen/qwen3-coder-480b-a35b-instruct)
[2026-04-29][00:06:53][INFO][cc_switch_lib::proxy::forwarder] [Claude] >>> 请求 URL: https://integrate.api.nvidia.com/v1/chat/completions (model=qwen/qwen3-coder-480b-a35b-instruct)
[2026-04-29][00:08:01][INFO][cc_switch_lib::proxy::forwarder] [Claude] >>> 请求 URL: https://integrate.api.nvidia.com/v1/chat/completions (model=qwen/qwen3-coder-480b-a35b-instruct)
nvidia一直是失败状态.
为什么会fallback到 主机 : api.anthropic.com:443?
1 个帖子 - 1 位参与者
来源: linux.do查看原文