Repository navigation
Custom provider with Openai compatible API Straico (successfully connected) showing blank output and keep looping #10573
Description
Description
- As stated in document https://opencode.ai/docs/providers/ . I want to create a custom provider connection to use with their API. In this case, it's Straico (The API docs here)
- It successful (api connection) but the output of the OpenCode seems frozen, but actually it's looping something (as my API credit get spent nonstop)
- This issue seems like happening with other users as well, but in different situation with different provider. (search in Github and Discord)
- My custom provider setting in config.json file.
"$schema": "https://opencode.ai/config.json",
"provider": {
"straico": {
"npm": "@ai-sdk/openai-compatible",
"name": "Straico",
"options": {
"baseURL": "https://api.straico.com/v2"
},
"models": {
"anthropic/claude-sonnet-4.5": {
"name": "Claude 4.5 Sonnet"
},
"openai/gpt-5.2": {
"name": "GPT-5.2"
},
"x-ai/grok-4-fast-reasoning": {
"name": "Grok 4 Fast Reasoning"
}
}
}
}
Plugins
MCP, Chrome DevTool
OpenCode version
1.1.30 until 1.1.35
Steps to reproduce
No response
Screenshot and/or share link
Operating System
MacOS Tahoe 26.2
Terminal
Warp
Metadata
Metadata
Assignees
Labels
No labels
Activity
github-actions commentedon Jan 25, 2026
This issue might be a duplicate of existing issues. Please check:
Feel free to ignore if none of these address your specific case.
mosfor commentedon Jan 25, 2026
yeah I have the same issue and without a custom provider just with the default provider and also with other providers like google, for me the first request goes will but some times the second or the third request wont output anything just looping
abankalarm commentedon Feb 11, 2026
Same issue, OpenAi custom provider calls fails and loop, antrhropic provider bugs out sometimes but works.
In network logs, i can see openai provider trying to send the same user request in loop
qwowboyp commentedon Feb 22, 2026
I've also encountered this using the built‑in xai/grok-4-1-fast. I started a small TODO task and it ran normally at first. I stepped away for about two hours, and when I came back the UI was blank with an "interrupt" indicator in the lower-left; the TODO had only reached about 17% completion. Checking the xAI dashboard shows about $9 of usage was consumed, but the OpenCode sidebar context shows less than $0.10.
As a note: pressing the ESC key twice within the same session seemed to restore normal operation. I switched to Gemini 3.1 Pro and it completed the remaining work in 2 minutes 45 seconds. I am using OpenCode version 1.2.10.
fix: use custom OpenAI-compatible provider to resolve empty responses
Delqhi commentedon Mar 7, 2026
This looks like the same custom-provider failure family we hit: the provider connects, but OpenCode ends up with blank output or a broken loop because the endpoint only partially matches the OpenAI tool / stream contract.
What worked in our fork was an opt-in compatibility layer for custom providers:
tools/tool_choice-> legacyfunctions/function_callfunction_call, text-only tool intents, JSON tool blocks, and wrapped<tool_call>payloadsfinish_reason=stopThat made a previously broken OpenAI-compatible provider usable for multi-turn tool execution on our side.
Related issues: #234, #15756, #5210.
If the maintainers are interested, I can turn that into a PR behind explicit custom-provider config.
github-actions commentedon Jul 6, 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.