Parameter max-completion-tokens is not working

[woeterman94]

I'm using

  - name: Generate human-readable report
        id: inference
        uses: actions/ai-inference@v2
        with:
          model: openai/gpt-5
          max-completion-tokens: 4000
          prompt: |
            Turn this .... deployment what-if JSON into a concise human-readable Markdown report.

And the output gives me

endpoint: https://models.github.ai/inference
    system-prompt: You are a helpful assistant
    max-tokens: 200
    token: ***
    enable-github-mcp: false
    provider: github-models
Using legacy prompt format

Max tokens is set to 200 ? Why? Is this max-completion-tokens parameter ignored?

The docs state
"(deprecated, use max-completion-tokens instead)"