What this error means

API error 400: The provider does not support the prompt-caching-scope-2026-01-05 beta header is a Claude Code failure pattern reported for developers trying to fix claude code sending prompt-caching beta header to custom api endpoints (anthropic_base_url) when caching is disabled. Based on the imported evidence, treat this as a tool-specific troubleshooting page rather than a generic API error.

Why this happens

GitHub issue #21644 in anthropics/claude-code: With enablePromptCaching=false and useAnthropicBetaHeaders=false set in VS Code settings.json, and a custom ANTHROPIC_BASE_URL that doesn't support prompt-caching-scope-2026-01-05, the beta header is still sent, causing API errors when using claude-4.5-opus model.

Common causes

  • Even when users explicitly disable prompt caching in VS Code settings (claudeCode.enablePromptCaching: false) and unset beta headers (claudeCode.useAnthropicBetaHeaders: false), Claude Code v2.1.23 still sends the 'prompt-caching-scope-2026-01-05' beta header to custom API endpoints. This breaks integration with self-hosted or proxy API servers that don't support the beta.
  • GitHub issue #21644 in anthropics/claude-code: With enablePromptCaching=false and useAnthropicBetaHeaders=false set in VS Code settings.json, and a custom ANTHROPIC_BASE_URL that doesn't support prompt-caching-scope-2026-01-05, the beta header is still sent, causing API errors when using claude-4.5-opus model.

Quick fixes

  1. Confirm the exact error signature matches API error 400: The provider does not support the prompt-caching-scope-2026-01-05 beta header.
  2. Check the Claude Code account, local tool state, and provider configuration involved in the failing workflow.
  3. Compare the failing environment with a known working setup, then change one configuration value at a time.

Platform/tool-specific checks

  • Verify the command, editor, extension, or API client that produced the error.
  • Compare local settings with CI, deployment, or editor-level settings when the error appears in only one environment.
  • Avoid deleting credentials, local model data, or project settings until the failing scope is clear.

Step-by-step troubleshooting

  1. Capture the exact error message and the command, editor action, or request that triggered it.
  2. Check whether the failure is account/auth, quota/rate, model/provider, local runtime, or deployment configuration.
  3. Review the source evidence below and compare it with your environment.
  4. Apply one change at a time and rerun the smallest failing action.
  5. Keep the working fix documented for the team or deployment environment.

How to prevent it

  • Keep provider/tool configuration documented.
  • Record non-secret diagnostics such as tool version, provider name, model name, and command path.
  • Add a lightweight check before CI or production workflows depend on the tool.