Repository navigation
Inference is temporarily unavailable #34893
Description
Description
Inference is temporarily unavailable for Last 5 minutes on Ubuntu Machine
Model: Deepseek v4 Flash (Opencode Go)
Plugins
No response
OpenCode version
1.17.13
Steps to reproduce
No response
Screenshot and/or share link
No response
Operating System
Ubuntu 24.04
Terminal
No response
Metadata
Metadata
Assignees
Labels
No labels
Activity
BlackC1over commentedon Jul 2, 2026
same problem in windows
model: DeepSeek v4 flash
lordrebel commentedon Jul 2, 2026
me too!
my log is here:
model=opencodego:deepseek-v4-flash::session-2026-05-21-b rawModel=deepseek-v4-flash endpoint=chat-completions metadataSource=models.dev messages=5 session=vscode-05834fbd request=req-0e932c15 modelConfiguration={"reasoningEffort":"off"} thinking={"deepseek":"off","glm":"off","kimi":"off","minimax":"off","qwen":"off","qwenBudget":"auto","mimo":"off"} thinkingPayload={}
[request] url=https://opencode.ai/zen/go/v1/chat/completions payloadBytes=122796 requestTimeoutMs=600000 streamIdleTimeoutMs=120000
[http] 503 Service Unavailable content-type=application/json
[http-error-body] {"error":{"message":"Inference is temporarily unavailable","type":"server_error","param":null,"code":"failover_exhausted"}}
[response-summary] status=503 durationMs=26280 ttfbMs=n/a promptTokens=n/a completionTokens=n/a totalTokens=n/a cachedTokens=n/a finishReason= totalBytes=123 totalEvents=0
[2026-07-02T09:20:37.727Z] [go-usage] Recording: provider=OpenCode Go model=deepseek-v4-flash promptTokens=n/a completionTokens=n/a cachedTokens=n/a status=503 error=OpenCode Go API request failed (503) model=deepseek-v4-flash payloadBytes=122796: Inference is temporarily unavailable
[2026-07-02T09:20:37.728Z] [go-usage] After record: entries=0
[2026-07-02T09:20:37.729Z] ERROR model=opencodego:deepseek-v4-flash::session-2026-05-21-b: OpenCode Go API request failed (503) model=deepseek-v4-flash payloadBytes=122796: Inference is temporarily unavailable
but opecode zen can work
github-actions commentedon Jul 2, 2026
Thanks for updating the issue! It now has the key details needed (model, provider, OS, and version). We appreciate you taking the time to fill things out.
smok95 commentedon Jul 2, 2026
API call failed (attempt 1/3): InternalServerError [HTTP 503]
⚠️ API call failed (attempt 2/3): InternalServerError [HTTP 503]
⚠️ API call failed (attempt 3/3): InternalServerError [HTTP 503]
🔌 Provider: opencode-go Model: deepseek-v4-flash
🌐 Endpoint: https://opencode.ai/zen/go/v1
📝 Error: HTTP 503: Inference is temporarily unavailable
⏱️ Elapsed: 24.89s Context: 2 msgs, ~7,710 tokens
⏳ Retrying in 2.7s (attempt 1/3)...
🔌 Provider: opencode-go Model: deepseek-v4-flash
🌐 Endpoint: https://opencode.ai/zen/go/v1
📝 Error: HTTP 503: Inference is temporarily unavailable
⏱️ Elapsed: 52.48s Context: 2 msgs, ~7,710 tokens
⏳ Retrying in 5.4s (attempt 2/3)...
🔌 Provider: opencode-go Model: deepseek-v4-flash
🌐 Endpoint: https://opencode.ai/zen/go/v1
📝 Error: HTTP 503: Inference is temporarily unavailable
⏱️ Elapsed: 82.80s Context: 2 msgs, ~7,710 tokens
❌ API failed after 3 retries — HTTP 503: Inference is temporarily unavailable
💀 Final error: HTTP 503: Inference is temporarily unavailable
GGSDDUAY commentedon Jul 2, 2026
me too!
Johell1NS commentedon Jul 2, 2026
me too
SamMousa commentedon Jul 2, 2026
Switching to another model supported by opencode go works.
philiaspaceai commentedon Jul 2, 2026
me too bro, we have the same problem here
ArnChowdhury commentedon Jul 2, 2026
us bro us
TaotaoZLG commentedon Jul 2, 2026
+1 bro
gianghp123 commentedon Jul 2, 2026
same for mimo-v2.5
Raingor commentedon Jul 2, 2026
me too
11 remaining items
roychanhy commentedon Jul 2, 2026
all dead
🔌 Provider: opencode-go Model: qwen3.6-plus
🌐 Endpoint: https://opencode.ai/zen/go/v1
📝 Error: HTTP 503: Inference is temporarily unavailable
⏱️ Elapsed: 1.75s Context: 2 msgs, ~6,397 tokens
⏳ Retrying in 2.6s (attempt 1/3)...
🔌 Provider: opencode-go Model: qwen3.6-plus
🌐 Endpoint: https://opencode.ai/zen/go/v1
📝 Error: HTTP 503: Inference is temporarily unavailable
⏱️ Elapsed: 5.55s Context: 2 msgs, ~6,397 tokens
⏳ Retrying in 4.8s (attempt 2/3)...
🔌 Provider: opencode-go Model: qwen3.6-plus
🌐 Endpoint: https://opencode.ai/zen/go/v1
📝 Error: HTTP 503: Inference is temporarily unavailable
⏱️ Elapsed: 11.83s Context: 2 msgs, ~6,397 tokens
❌ API failed after 3 retries — HTTP 503: Inference is temporarily unavailable
💀 Final error: HTTP 503: Inference is temporarily unavailable
Amouren7 commentedon Jul 2, 2026
deepseek和mimo都是这样，我试了一下minmiax都没问题
Arnav-Suraj-Nagesh commentedon Jul 2, 2026
I am facing the same issue. Please fix it at the earliest.
Raingor commentedon Jul 2, 2026
其他模型可以吗
k176060444-lgtm commentedon Jul 2, 2026
me too too
z454963058-sudo commentedon Jul 2, 2026
me too
LouiseValliere commentedon Jul 2, 2026
same
ishaan-jindal commentedon Jul 2, 2026
seems to be working now
lordrebel commentedon Jul 2, 2026
The outage seems to have resolved on its own — DeepSeek V4 Flash is working again on my end. Thanks for getting it sorted!
Just a gentle suggestion: it would be really helpful if the team could drop a quick note next time when something like this happens — even a short "we are aware, working on it" during the incident or a "should be fixed now" afterwards. It saves a lot of users from wondering whether it is their config, their account, or a global issue.
Appreciate all the work you do on OpenCode! 🙏
lordrebel commentedon Jul 2, 2026
This comment was drafted by AI and posted with the user's authorization.
Raingor commentedon Jul 2, 2026
iit work? but i connect time out
ishaan-jindal commentedon Jul 2, 2026
Is it down again? Since I am experiencing outages again...
1105623876 commentedon Jul 3, 2026
all models from qwen failed from yesterday
fwang commentedon Jul 4, 2026
we are migrating our inference servers .. things are already much more stable
thanks for your patience