mirror of
https://github.com/deepseek-ai/awesome-deepseek-agent.git
synced 2026-11-05 01:07:37 +00:00
[GH-ISSUE #62] claude desktop接入不了 #18
Labels
No labels
pull-request
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
deepseek-ai/awesome-deepseek-agent#18
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @lxy2020 on GitHub (May 6, 2026).
Original GitHub issue: https://github.com/deepseek-ai/awesome-deepseek-agent/issues/62
用claude code的地址接不了claude desktop，会报错。
报错信息：
message: Gateway returned HTTP 404
httpStatus: 404
requestUrl: https://api.deepseek.com/anthropic/v1/models
endpoint: https://api.deepseek.com/
checkedAt: 2026-05-06T03:38:56.651Z
@lxy2020 commented on GitHub (May 6, 2026):
配置了 Model 还是会报错。
报错信息：
message: Invalid custom3p enterprise config: inferenceModels: configured model "deepseek-v4-flash" is not an Anthropic model. Gateway deployments require an Anthropic model from the provider catalog — expected a gateway model route referencing an Anthropic model (e.g. claude-sonnet-4-5, anthropic/claude-*). Name routes to match the underlying model.
failingField: inferenceModels
checkedAt: 2026-05-06T03:44:16.771Z
@lxy2020 commented on GitHub (May 6, 2026):
window 11
@liuxinhh commented on GitHub (May 6, 2026):
在下方添加下自定义大模型列表, 就可以了
@lxy2020 commented on GitHub (May 6, 2026):
配了，还是报错
message: Invalid custom3p enterprise config: inferenceModels: configured model "deepseek-v4-flash" is not an Anthropic model. Gateway deployments require an Anthropic model from the provider catalog — expected a gateway model route referencing an Anthropic model (e.g. claude-sonnet-4-5, anthropic/claude-*). Name routes to match the underlying model.
failingField: inferenceModels
checkedAt: 2026-05-06T03:44:16.771Z
@liuxinhh commented on GitHub (May 6, 2026):
@liuxinhh commented on GitHub (May 6, 2026):
我的版本是 Claude 1.4758.0 (fb266c) 2026-04-24T20:22:30.000Z
@jiangxueqians-ai commented on GitHub (May 6, 2026):
版本 1.6259.1 (5095e7)测试联接同样错误，应该是claude自己封的。
@Ltb2539 commented on GitHub (May 6, 2026):
可以试试这个 https://github.com/Ltb2539/UpstreamKit 刚用codex写的工具 实测Claude 1.6259.0 (dc89db) 2026-05-05T18:14:20.000Z 可以运行
@Agents365-ai commented on GitHub (May 6, 2026):
Try https://github.com/Agents365-ai/dsclaude
one script to set claude dekstop, and use co-work, claude code
@acewong7 commented on GitHub (May 6, 2026):
version 1.5354.0 可以直接配deepseek-v4-pro和deepseek-v4-flash模型，升级到1.6259.1后不行了，需要在配置里模型前面加claude-才行，但是问题是claude-deepseek-v4-pro的模型传给deepseek后可能由于模型名字不对自动降级到备用的deepseek-v4-flash模型，claude desktop还是用不到pro模型？看论坛有人建议写个http proxy把模型名字再换回官方的deepseek-v4-pro。https://www.reddit.com/r/LocalLLM/comments/1t3oypq/switched_my_claude_code_agent_loop_to_deepseek_v4/
@TSG9876 commented on GitHub (May 11, 2026):
use cc-Switch to bridge to actual deepseek api end point , enable cc-Switch local routing as default http://127.0.01:15721, configure your cc Switch claude end point to actual deepseek api endpoint https://api.deepseek.com/anthropic
, and the actual model deepseek id
in the claude code , set model id as per suggestion with prefix claude-your model , gateway set as http://127.0.01:15721
@Agents365-ai commented on GitHub (May 11, 2026):
不需要proxy
Try https://github.com/Agents365-ai/dsclaude
one script to set claude dekstop, and use co-work, claude code
@lxy2020 commented on GitHub (May 13, 2026):
找到了个新解决方案，B站上看到的
在环境变量里面配置一下就好了
环境变量名：NODE_ENV 变量值：production
@lxy2020 commented on GitHub (May 19, 2026):
1.7 之后又有新的限制，这个方法失效了
@BiBi2644 commented on GitHub (May 27, 2026):
怎么关闭自动更新呀
@denial123789 commented on GitHub (Aug 28, 2026):
补充一个不修改 Claude Desktop 第三方模型配置的替代路径：可以使用 SandBase CLI 的本地 MCP 桥接，让支持 MCP 的客户端通过 OAuth 发现并调用 DeepSeek 模型/API。
这不是对 Claude Desktop
anthropic/v1/models404 的直接修复；它绕开了该 Provider 端点，适合希望继续使用 MCP 客户端能力的场景。仓库和官方 Registry：https://github.com/sandbaseai/cli
https://registry.modelcontextprotocol.io/v0.1/servers/io.github.sandbaseai%2Fcli
Disclosure: 我代表 SandBase AI 提供这个替代方案，具体兼容性请按客户端版本验证。