Skip to content

Truncated / wrong prefix & suffix on large file Autocompletion #12390

@ArifBS

Description

@ArifBS

Before submitting your bug report

Relevant environment info

- OS: Linux Ubuntu
- Continue version: 1.0.67
- IDE version: Intellij Idea 2026.1.1
- Model: Deepseek v4 flash & codestral
- config:
  

  - name: DeepSeek Autocompletion
    provider: deepseek
    model: deepseek-v4-flash
    apiBase: https://api.deepseek.com
    apiKey: sk-xxxxx
    defaultCompletionOptions:
      contextLength: 128000
      maxTokens: 256
    autocompleteOptions:
      debounceDelay: 350
      maxPromptTokens: 8192
      prefixPercentage: 0.5
      maxSuffixPercentage: 0.5
      multilineCompletions: true


  
  OR link to agent in Continue hub:

Description

Hello,

The autocompletion is working fine on small file (tens or hundreds of lines) but failing on larger file (more than 3000 lines).
On the larger file, it show suggestion that totally wrong and when I move the cursor into another place it still show that same wrong suggestion.

Digging the prompt.log I found that the prefix and suffix are incorrect. Doesn't matter where the cursor is, it get the same wrong prefix & suffix.

here is the log on small and large file.

prompt_correct.log
prompt_wrong.log

Note:

  • The vscode plugin works fine using the same yaml configuration.

thanks
ArifBS

To reproduce

No response

Log output

Metadata

Metadata

Assignees

No one assigned

    Labels

    area:autocompleteRelates to the auto complete featureide:jetbrainsRelates specifically to JetBrains extensionkind:bugIndicates an unexpected problem or unintended behavioros:linuxHappening specifically on Linux

    Type

    No type

    Projects

    Status

    Todo

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions