No tool call found for function call output with call_id

[rkocri]

Before submitting your bug report

• I've tried using the "Ask AI" feature on the Continue docs site to see if the docs have an answer
• I'm not able to find a related conversation on GitHub discussions that reports the same bug
• I'm not able to find an open issue that reports the same bug
• I've seen the troubleshooting guide on the Continue Docs

Relevant environment info

- OS:win11
- Continue version: 10.0.52
- IDE version: IntelliJ IDEA 2025.2.4 (Ultimate Edition)
- Model: GPT-4.1
- config:
  
name: Local Assistant
version: 1.0.0
schema: v1
models:
  - uses: openai/gpt-5
    with:
      OPENAI_API_KEY: sk-proj-XXX
  - uses: openai/o3
    with:
      OPENAI_API_KEY: sk-proj-XXX
  - uses: openai/o3
    with:
      OPENAI_API_KEY: sk-proj-XXX
  - uses: openai/gpt-4.1-mini
    with:
      OPENAI_API_KEY: sk-proj-XXX
  - name: Autodetect
    provider: openai
    model: AUTODETECT
    requestOptions:
      timeout: 60000  # Request timeout in milliseconds
      verifySsl: true
    apiKey: sk-proj-XXX
context:
  - provider: code
  - provider: docs
  - provider: diff
  - provider: terminal
  - provider: problems
  - provider: folder
  - provider: codebase

  
  OR link to agent in Continue hub:

Description

When working with continue and all tools activated I allways get this error after some time.
Sometimes it helps to resubmit last message or edit last prompt to start generating from there, but
most sessions do not recover from that and I have to live with half-finished tasks an have to start a new session,
which is extremely annoying and time consuming.

Error-output:
{"message":"No tool call found for function call output with call_id call_ubrVpxVSFon3MNAk8df8F0kO.","type":"invalid_request_error","param":"input","code":null}

To reproduce

No response

Log output

{
  "error": {
    "message": "No tool call found for function call output with call_id call_ubrVpxVSFon3MNAk8df8F0kO.",
    "type": "invalid_request_error",
    "param": "input",
    "code": null
  }
}

Code: undefined
Error number: undefined
Syscall: undefined
Type: undefined

Error: HTTP 400 Bad Request from https://api.openai.com/v1/responses

{
  "error": {
    "message": "No tool call found for function call output with call_id call_ubrVpxVSFon3MNAk8df8F0kO.",
    "type": "invalid_request_error",
    "param": "input",
    "code": null
  }
}
    at aEt.parseError (C:\snapshot\continue\binary\out\index.js:8421:15852)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async n (C:\snapshot\continue\binary\out\index.js:8423:144)
    at async Fht (C:\snapshot\continue\binary\out\index.js:8161:3322)
    at async aEt._streamResponses (C:\snapshot\continue\binary\out\index.js:8444:7591)
    at async aEt.responsesStream (C:\snapshot\continue\binary\out\index.js:8438:4965)
    at async aEt.streamChat (C:\snapshot\continue\binary\out\index.js:8438:6480)
    at async G0a (C:\snapshot\continue\binary\out\index.js:9631:2628)
    at async C:\snapshot\continue\binary\out\index.js:9690:6927 
[2025-11-18T16:38:40] Error: HTTP 400 Bad Request from https://api.openai.com/v1/responses

{
  "error": {
    "message": "No tool call found for function call output with call_id call_ubrVpxVSFon3MNAk8df8F0kO.",
    "type": "invalid_request_error",
    "param": "input",
    "code": null
  }
} 
[2025-11-18T16:38:40] Error running handler for "llm/streamChat":  Error: HTTP 400 Bad Request from https://api.openai.com/v1/responses

{
  "error": {
    "message": "No tool call found for function call output with call_id call_ubrVpxVSFon3MNAk8df8F0kO.",
    "type": "invalid_request_error",
    "param": "input",
    "code": null
  }
}

[Quazistax]

It happens when you start a chat, do some tool calls (e.g. ask AI to run terminal command), exit VS code, start VS code again, try to continue with chat.
When history/messages are read from old session files on VS code startup, they do not have all the data (response id) as when they are created in memory by chat request/responses. Can be solved by ensuring unique item id is specified/generated for old function_call messages. This fixed worked for me: fix(openai): responses function_call mapping

[houssemzaier]

I've been hitting this issue with GPT-5/GPT-5.1 using the Responses API and parallel tool calls.

Opened PR #9659 with a fix. The root cause I found: when the model makes parallel tool calls, each function_call has a unique call_id (fc_*), but these IDs were being overwritten instead of accumulated during streaming.

My fix accumulates the IDs in a responsesOutputItemIds[] array and matches them positionally when building the next request.

Note: This fix addresses the streaming/parallel tool calls case. The legacy session issue mentioned above might need a separate fix (generating IDs for old sessions without them).

Tested with GPT-5, GPT-5.1, and GPT-5 Codex.

[sestinj]

🎉 This issue has been resolved in version 1.41.0 🎉

The release is available on:

• npm package (@latest dist-tag)
• GitHub release

Your semantic-release bot 📦🚀