[BUG] weird todo/tool call issues cause the model to stop generating

[cau1k]

Bug Description
Wondering if this a bug with the plugin. i only get this with gpt models via the codex plugin. Never happened before the latest update

Steps to Reproduce

1. Run gpt-5.2-codex:xhigh in a loop/long task
2. Emphasize use of todos

Expected Behavior
To not stop generating after listing todos

Actual Behavior

Environment

• opencode version: latest
• Plugin version: latest
• Operating System: linux
• Node.js version: 24 LTS

Logs
If applicable, attach logs from ~/.opencode/logs/codex-plugin/ (enable with ENABLE_PLUGIN_REQUEST_LOGGING=1)

Compliance Checklist
Please confirm:

• I'm using this plugin for personal development only
• I have an active ChatGPT Plus/Pro subscription
• This issue is not related to attempting commercial use or TOS violations
• I've reviewed the FAQ and Troubleshooting sections

Additional Context
Add any other relevant information.

[cau1k]

@claude investigate this habibi

[CruelAddict]

Having this issue too. Also, the models got extremely dumb, it's barely possible to squeeze any file edits out of them after updating to the new config. For example, during the session from this screenshot it was "planning" the same edit for about 5 minutes, while making the same 1-line edit 5 times

Switching back to the legacy version seems to have made it back to normal, although my openai sessions have no titles now

[plum-dawg]

also having this issue.

[cau1k]

@plum-dawg @CruelAddict are y'all using any other plugins?

opencode released some semi-breaking changes in 1.1.0 which could be causing this. For reference, DCP kinda broke with the release of variants (see Opencode-DCP/opencode-dynamic-context-pruning#225)

[mike2153]

im having this same issue as well. Some chats responses will only run 5 seconds and respond weirdly as well completely unrelated to to the chat. Also tools are causing the model to stop completely

[reginthomsen]

Also having this issue.

[OliverBuS]

Hey. I just had this issue a lot. I checked my logs and found the following:

ERROR 2026-01-08T09:00:33 +253ms service=llm providerID=openai modelID=gpt-5-nano sessionID=ses_463334694ffe2I5S2WhxYDaAEn small=true agent=title error={"error":{"name":"AI_APICallError","url":"https://chatgpt.com/backend-api/responses","requestBodyValues":{"model":"gpt-5-nano","input":[{"role":"developer","content":"You are a title generator. You output ONLY a thread title. Nothing else.\n\n<task>\nGenerate a brief title that would help the user find this conversation later.\n\nFollow all rules in <rules>\nUse the <examples> so you know what a good title looks like.\nYour output must be:\n- A single line\n- ≤50 characters\n- No explanations\n</task>\n\n<rules>\n- Focus on the main topic or question the user needs to retrieve\n- Use -ing verbs for actions (Debugging, Implementing, Analyzing)\n- Keep exact: technical terms, numbers, filenames, HTTP codes\n- Remove: the, this, my, a, an\n- Never assume tech stack\n- Never use tools\n- NEVER respond to questions, just generate a title for the conversation\n- The title should NEVER include \"summarizing\" or \"generating\" when generating a title\n- DO NOT SAY YOU CANNOT GENERATE A TITLE OR COMPLAIN ABOUT THE INPUT\n- Always output something meaningful, even if the input is minimal.\n- If the user message is short or conversational (e.g. \"hello\", \"lol\", \"what's up\", \"hey\"):\n  → create a title that reflects the user's tone or intent (such as Greeting, Quick check-in, Light chat, Intro message, etc.)\n</rules>\n\n<examples>\n\"debug 500 errors in production\" → Debugging production 500 errors\n\"refactor user service\" → Refactoring user service\n\"why is app.js failing\" → Analyzing app.js failure\n\"implement rate limiting\" → Implementing rate limiting\n\"how do I connect postgres to my API\" → Connecting Postgres to API\n\"best practices for React hooks\" → React hooks best practices\n</examples>\n"},{"role":"user","content":[{"type":"input_text","text":"\n              The following is the text to summarize:\n              <text>\n              Again?...\n              </text>\n            "}]}],"max_output_tokens":32000,"prompt_cache_key":"ses_463334694ffe2I5S2WhxYDaAEn","reasoning":{"effort":"minimal"},"stream":true},"statusCode":400,"responseHeaders":{"cf-cache-status":"DYNAMIC","cf-ray":"9baa8156fa9a6def-LIM","connection":"keep-alive","content-length":"271","cross-origin-opener-policy":"same-origin-allow-popups","date":"Thu, 08 Jan 2026 09:00:33 GMT","nel":"{\"success_fraction\":0.01,\"report_to\":\"cf-nel\",\"max_age\":604800}","referrer-policy":"strict-origin-when-cross-origin","report-to":"{\"endpoints\":[{\"url\":\"https:\\/\\/a.nel.cloudflare.com\\/report\\/v4?s=dFfWGmMrmRM2POr924vaaKAbvZObqpGX0IIKdhv5bF7Y11N99vyYxRyGkV0EaKvrqViIGvDVX7XSyRT3umexrMQ9aTe9BIqC7OTQSeBlSMHV0TXGlaC2X%2FjHYKs7NJyVAQoZDtWlk31G\"}],\"group\":\"cf-nel\",\"max_age\":604800}","server":"cloudflare","strict-transport-security":"max-age=31536000; includeSubDomains; preload","x-content-type-options":"nosniff","x-envoy-upstream-service-time":"93","x-oai-request-id":"38fd6384-2e83-488f-9739-da6ca8f43114","x-openai-proxy-wasm":"v0.1","set-cookie":"__cf_bm=1OE03GOiwFmAFQi_DR.LBUf_Ti2DqXJDGhqGuIvT8UM-1767862833-1.0.1.1-NOFd15z.ds8wloGbalU9sPxxdELosBV58ue8GBXBF33V8UhLteiB30BxF.GyQTCOGDEHnYNdcGuufv2VFB5hW9a8D56s5Gwqsm2dcx52cXE; path=/; expires=Thu, 08-Jan-26 09:30:33 GMT; domain=.chatgpt.com; HttpOnly; Secure; SameSite=None"},"responseBody":"{\n  \"error\": {\n    \"message\": \"Unsupported value: 'minimal' is not supported with the 'gpt-5.1' model. Supported values are: 'none', 'low', 'medium', and 'high'.\",\n    \"type\": \"invalid_request_error\",\n    \"param\": \"reasoning.effort\",\n    \"code\": \"unsupported_value\"\n  }\n}","isRetryable":false,"data":{"error":{"message":"Unsupported value: 'minimal' is not supported with the 'gpt-5.1' model. Supported values are: 'none', 'low', 'medium', and 'high'.","type":"invalid_request_error","param":"reasoning.effort","code":"unsupported_value"}}}} stream error
ERROR 2026-01-08T09:00:33 +0ms service=default e=No output generated. Check the stream for errors. rejection

It seems it's calling the OpenAI API with minimal reasoning for the gpt-5-nano model (which is used to generate the session title). However, minimal is no longer a valid parameter; the API expects none, low, medium, or high.

The issue is located in lib/request/request-transformer.ts inside the getReasoningConfig function. It incorrectly assigns minimal to lightweight models (including nano)

I believe to fix it the logic needs to be updated to use low instead of minimal .

But I dont know why it stops the stream response, perhaps it could fix the sessions having no title assigned

[cau1k]

Yeah I'm back on my initial PR that was actually tested by a human before being merged

#91

git clone https://github.com/numman-ali/opencode-openai-codex-auth.git

cd opencode-openai-codex-auth

gh pr checkout 91

npm install

npm run build

Then update opencode.jsonc with the absolute path to this repo. Might have to add '/dist'

[numman-ali]

@cau1k that's quite an unneeded take. Seriously, like you've used the plug-in. I've put more than 30 hours into this plug-in making it work for everyone. Spent time as much as I can, and that's your take on it? Like it's pretty disrespectful.

I thought it would be funny to let the agent put out the joke that it did the work, but I did test everything fully to the best that I could through every single run, through every single model, tool calls, etc. There have been a few mistakes. That's normal, and that's okay, but come on, like this is free, you're not paying for it. I'd appreciate if you can be a bit more considerate rather than coming out with responses like that.

[plum-dawg]

@numman-ali no other pluggies

[iamhenry]

im also running into this error when it tries to use google search tool or the native webfetch from opencode

[openai-codex-plugin] 400 error: {"error":{"message":"No tool call found for fun
ction call output with call_id call_1526aS89sdj90BzWdJ4.","type":"invalid_request_error","param":"input","code":null,"status":400}}

GPT 5.2 Status: Failed 3x — google_search tool calls succeed but response processing errors out.

[numman-ali]

I'm fixing it today

[numman-ali]

Try the update

npx -y opencode-openai-codex-auth@latest

[numman-ali]

Fixed in v4.3.1: reasoning now normalizes to for GPT‑5.x, preventing backend rejections and stream/tool failures.\n\nUpdate: Backup created: /Users/numman/.config/opencode/opencode.json.bak-2026-01-09_14-01-27-344
Wrote /Users/numman/.config/opencode/opencode.json (modern config)

Done. Restart OpenCode to (re)install the plugin.
Example: opencode