Repository navigation
Bug Report #34903
Description
Description
Same issue here. DeepSeek V4 Flash and Pro via opencode-go have been broken since ~08:00 GMT today (July 2).
Evidence from our setup
Error from proxy logs (repeated since 08:00 GMT):
[request] url=https://opencode.ai/zen/go/v1/chat/completions
[http] 503 Service Unavailable
[http-error-body] {"error":{"message":"Inference is temporarily unavailable",
"type":"server_error","code":"failover_exhausted"}}
Direct curl test against https://opencode.ai/zen/go/v1/chat/completions:
- Previous API key returned:
403 error code: 1010(Cloudflare block) - After rotating to a fresh Go subscription API key →
503 Inference is temporarily unavailable - Both
deepseek-v4-flashanddeepseek-v4-proaffected
Other models via opencode-go still work? (e.g. GLM, Kimi)
Workaround confirmed: Direct DeepSeek API (https://api.deepseek.com with a separate key) works fine, so the issue is definitely on OpenCode's backend side, not DeepSeek's own API.
Setup
- Provider: opencode-go
- Model: deepseek-v4-flash / deepseek-v4-pro
- API endpoint: https://opencode.ai/zen/go/v1
- Go subscription: Active, fresh API key
Plugins
No response
OpenCode version
No response
Steps to reproduce
No response
Screenshot and/or share link
No response
Operating System
No response
Terminal
No response
Activity
github-actions commentedon Jul 2, 2026
This issue might be a duplicate of existing issues. Please check:
These appear to be part of the same backend outage affecting DeepSeek models on opencode-go since ~08:00 GMT July 2.
ruslankonev commentedon Jul 2, 2026
Opencode Go plan also doesn't work. It say – Error from provider (Console Go): Provider rate limit exceeded
but that's not true
github-actions commentedon Sep 1, 2026
To stay organized issues are automatically closed after 60 days of no activity. If the issue is still relevant please open a new one.