{"version":"2.1.288","anchor":"new-env-var-claude-code-disable-structured-outputs-turns-off","canonical_anchor":"new-env-var-claude-code-disable-structured-outputs-turns-off","heading":"CLAUDE_CODE_DISABLE_STRUCTURED_OUTPUTS turns off structured outputs","tier":"use","area":"Structured Outputs","scope":"individual","heads_up":false,"url":"https:\/\/changelogs.core-directive.com\/v\/2.1.288\/e\/new-env-var-claude-code-disable-structured-outputs-turns-off","release_url":"https:\/\/changelogs.core-directive.com\/v\/2.1.288","markdown":"### `CLAUDE_CODE_DISABLE_STRUCTURED_OUTPUTS` turns off structured outputs\n\nSetting `CLAUDE_CODE_DISABLE_STRUCTURED_OUTPUTS` stops Claude Code from asking the model for replies in a fixed format\n\n**Unclear.** It is not clear which models the new support check excludes.\n\n**What**\n\nStructured outputs are requests where Claude Code asks the model to reply in a fixed, machine-readable format. You can now turn them off by setting the environment variable `CLAUDE_CODE_DISABLE_STRUCTURED_OUTPUTS`. An environment variable is a named value you set in your shell before starting Claude Code.\n\nThe step that applies a reply format also skips models that a separate check marks as unsupported. Before this change, only one model was excluded and there was no way to opt out.\n\n**Why**\n\nIf your gateway or proxy (a service that sits between Claude Code and the API) rejects structured-output requests, this gives you a way to switch them off.\n\n- Area: Structured Outputs\n- Names: `CLAUDE_CODE_DISABLE_STRUCTURED_OUTPUTS`\n- Tier: Use it now\n- Useful: 5\/5\n- Signal: 3\/5\n- Scope: individual\n- Heads-up: no"}