DeepSeek v4.1 Flash always returns 400 #54117

Activity

1. github-actions

   assigned

   vaprdev

   on Oct 9, 2026
2. opencode-agent commented on Oct 9, 2026

   

   opencode-agentBot

   on Oct 9, 2026 – with OpenCode Agent

   Contributor

   More actions

   Thanks for the report. I couldn't reproduce this here: the 400 comes from the OpenCode Go service, and our test sandbox has no Go subscription, so a maintainer with access will need to look at it.

   Provider request failed with HTTP 400 is OpenCode's generic fallback for when the service's error reply doesn't say why it rejected the request, so the useful detail is missing. If you can, please add:

   • whether other Go models (e.g. DeepSeek V4 Pro, GLM) work for you in a new session
   • whether changing the reasoning variant/effort for DeepSeek V4.1 Flash makes any difference
   • any log lines from around the failure that show the response body

   These open issues are related, though in them the 400s come and go or appear late in long sessions, not on every prompt:

   • OpenCode Go DeepSeek V4.1 Flash intermittently returns HTTP 400 with only model name #51990: Go DeepSeek V4.1 Flash, intermittent 400 whose error body only names the model
   • opencode-go/deepseek-v4.1-flash randomly rejects requests with 400/403 #51434: random 400/403 from deepseek-v4.1-flash
   • Console Go relay: HTTP 400 (empty body) when request combines reasoning_content echo-back with reasoning_effort on deepseek-v4-flash #48180: 400 when reasoning effort is combined with replayed reasoning content
   • bug: OpenCode Go deepseek-flash returns 400 inference_failed on oversized payloads with max reasoning effort #52174: 400 on large requests with max reasoning effort
3. ww-TNT commented on Oct 16, 2026

   

   ww-TNT

   on Oct 16, 2026

   More actions

   

   I encountered HTTP 400 errors with both deepseek-v4-pro and deepseek-v4.1-flash on OpenCode Go.

   Workaround that fixed the issue:

   Open the OpenCode Go settings.

   Change Regions from Europe & United States to Global.

   After switching to Global, both models started working correctly.

   This suggests the issue may be related to regional routing or model availability rather than the models themselves.

   Hope this helps others experiencing the same issue.
4. neriousy commented on Oct 21, 2026

   

   neriousy

   on Oct 21, 2026

   Member

   More actions

   Thanks for the report. This is about how the model itself behaves when served through OpenCode Go (repeating itself, broken tool calls, speed, or rejecting large requests), not a bug in the OpenCode app. We pass these reports to our upstream providers. Starting a new session or running /compact usually helps with loops and oversized-request errors. Closing this to keep the tracker focused; the main discussion is in #50206.
5. neriousy

   closed this as not plannedon Oct 21, 2026