[BUG] `output_config` sent to Vertex AI causes `invalid_request_error` on session title generation

Status Fixed / completed
Reported on v2.1.118
Maintainer reply ✓ Yes — mhegazy
Activity 6 comments · opened Apr 23, 2026 · closed Apr 27, 2026
💡 Likely answer: A maintainer (mhegazy, contributor) responded on this thread — see the highlighted reply below.

Preflight Checklist

  • [x] I have searched existing issues and this hasn't been reported yet
  • [x] This is a single bug report (please file separate reports for different bugs)
  • [x] I am using the latest version of Claude Code

What's Wrong?

Environment

  • Claude Code version: 2.1.118 (Claude Code)
  • Backend: Vertex AI
  • Model: claude-haiku-4-5@20251001
  • Environment variables set:

``
CLAUDE_CODE_USE_VERTEX=1
CLOUD_ML_REGION=global
ANTHROPIC_VERTEX_PROJECT_ID=my-secret-project
VERTEX_REGION_CLAUDE_3_5_HAIKU=global
``

Describe the bug

When running Claude Code with Vertex AI, the session title generation request fails because Claude Code includes an output_config field in the request body. Vertex AI's Claude integration does not support output_config and rejects the request with an invalid_request_error.

Request body sent (abbreviated)

{
  "model": "claude-haiku-4-5@20251001",
  "output_config": {
    "format": {
      "type": "json_schema",
      "schema": {
        "type": "object",
        "properties": { "title": { "type": "string" } },
        "required": ["title"],
        "additionalProperties": false
      }
    }
  },
  ...
}

Error response from Vertex AI

{
  "type": "error",
  "error": {
    "type": "invalid_request_error",
    "message": "output_config: Extra inputs are not permitted"
  }
}

Actual behavior

The session title generation call fails. This appears to affect every new session start.

Workaround

None available for CLI users. Users proxying requests can strip output_config before forwarding to Vertex AI.

What Should Happen?

Expected behavior

When CLAUDE_CODE_USE_VERTEX=1 is set, Claude Code should not include output_config (or other Anthropic-native-only fields) in requests sent to Vertex AI. JSON output should be enforced via prompt instructions instead, which Vertex AI supports.

Error Messages/Logs

{
	"type": "error",
	"error": {
		"type": "invalid_request_error",
		"message": "output_config: Extra inputs are not permitted"
	},
	"request_id": "req_vrtx_..."
}

Steps to Reproduce

Start claude and send hi in a new session.

Claude Model

Other

Is this a regression?

I don't know

Last Working Version

_No response_

Claude Code Version

2.1.118 (Claude Code)

Platform

Google Vertex AI

Operating System

macOS

Terminal/Shell

Terminal.app (macOS)

Additional Information

_No response_

View original on GitHub ↗

6 Comments

github-actions[bot] · 4 months ago

Found 2 possible duplicate issues:

  1. https://github.com/anthropics/claude-code/issues/37746
  2. https://github.com/anthropics/claude-code/issues/41149

This issue will be automatically closed as a duplicate in 3 days.

  • If your issue is a duplicate, please close it and 👍 the existing issue instead
  • To prevent auto-closure, add a comment or 👎 this comment

🤖 Generated with Claude Code

aeroxy · 4 months ago

@mhegazy hey what's up? is this issue fixed?

mhegazy contributor · 4 months ago

Fix merged. should be in an upcoming release (likely v2.1.122).

aeroxy · 4 months ago

@mhegazy awesome! so fast! appreciate ur work

adrian-skybaker · 3 months ago

Same issue appears to occur for Bedrock - https://github.com/anthropics/claude-code/issues/57611

github-actions[bot] · 2 months ago

This issue has been automatically locked since it was closed and has not had any activity for 7 days. If you're experiencing a similar issue, please file a new issue and reference this one if it's relevant.