opencode go 的 deepseek v4 flash 也开始 429 了？

V2EX = way to explore

V2EX 是一个关于分享和探索的地方

For Existing Member Sign In

• 请不要在回答技术问题时复制粘贴 AI 生成的内容

，今天调用报 bad response status code 429, message: Error from provider (DeepSeek): Too many requests. Please pace your requests reasonably. Your current concurrency: 2000, body: {"error":{"message":"Error from provider (DeepSeek): Too many requests. Please pace your requests reasonably. Your current concurrency: 2000","type":"rate_limit_error","param":null,"code":"invalid_request_error"}}
我记得是采购自 deepseek 官方的，deepseek 官方的算力终于用满了，是不是要涨价了

rate_limit

DeepSeek

算力

13 replies • 2026-05-06 15:31:48 +08:00

knva

7h 30m ago

确实卡半天

lynn1su

7h 29m ago

@knva #1 万恶的 deepseek 官方

106npo

7h 18m ago

这是按用户限制了 2000 并发 ,opencode go 所有用户都视作同一用户了

lynn1su

7h 3m ago

@106npo #3 opencodego 估计在 deepseek 那就开了一个号，也不知道多开几个号轮询的

maocat

7h 0m ago via Android

@lynn1su 轮训缓存还有用吗

lynn1su

6h 59m ago

@maocat #5 有用的

zenfsharp

6h 11m ago

OpenCode 不是号称自建算力的么…

106npo

5h 43m ago

@zenfsharp 没见过文档里写的是我们测试了一组经过精选且与 OpenCode 配合良好的模型和提供商。

zhonghao01

5h 32m ago

@zenfsharp 他们文档里写了提供商，除了 GLM 有多个，其他全部是模型厂商的，没有一个是 opencode 提供的。

zenfsharp

5h 25m ago

@106npo
@zhonghao01
我看它说部署在欧洲、美国、新加坡，还以为开源的是自己部署的算力。

freekindom

5h 20m ago

换 pi

xocus

4h 19m ago

cli 不理 429 529 就可以了，遇到直接再请求就行了，我一直这么干的

PolarBears

3h 56m ago

是 deepseek 本身出问题的原因吧：
Past Incidents
May 6, 2026
[已恢复] DeepSeek 网页/API 性能异常（[Resolved]DeepSeek Web/API Service Degraded Performance ）
Resolved - This incident has been resolved.
May 6, 15:20 CST
Identified - The issue has been identified and a fix is being implemented.
May 6, 15:13 CST
[已恢复] DeepSeek 网页/API 性能异常（[Resolved]DeepSeek Web/API Service Degraded Performance ）
Resolved - This incident has been resolved.
May 6, 12:04 CST
Identified - The issue has been identified and a fix is being implemented.
May 6, 11:53 CST