33 ファイル変更+167-54
この更新の概要
最上位モデルClaude Fable 5の導入に伴い、モデル設定や自動フォールバック機能に関する詳細が追加されました。Planモードの挙動が変更され、ファイル編集の自動承認が行われず常にcanUseToolコールバックを介する仕組みになっています。フック機能では入出力内容を書き換えるための新しい制御プロパティが導入され、セキュリティ面ではコマンドインジェクション検出やネットワークコマンドの承認フローが強化されました。
Claude Fable 5およびMythos 5のスペック、API ID、トークナイザーの特性、および価格情報の詳細が追加されました。
@@ -6,17 +6,35 @@ Claude is a family of state-of-the-art large language models developed by Anthro
## Choosing a model
If you're unsure which model to use, consider starting with **Claude Opus 4.8** for the most complex tasks. It is Anthropic's most capable model for complex reasoning, long-horizon agentic coding, and high-autonomy work.
If you're unsure which model to use, consider starting with **Claude Opus 4.8** for the most complex tasks. It is Anthropic's most capable Opus-tier model for complex reasoning, long-horizon agentic coding, and high-autonomy work. For workloads that need the highest available capability, see [Claude Fable 5](#claude-fable-5-and-claude-mythos-5).
All current Claude models support text and image input, text output, multilingual capabilities, and vision. Models are available through the Claude API, [Claude Platform on AWS](/docs/en/build-with-claude/claude-platform-on-aws), [Amazon Bedrock](/docs/en/build-with-claude/claude-in-amazon-bedrock), [Vertex AI](/docs/en/build-with-claude/claude-on-vertex-ai), and [Microsoft Foundry](/docs/en/build-with-claude/claude-in-microsoft-foundry).
Once you've picked a model, [learn how to make your first API call](/docs/en/get-started).
### Claude Fable 5 and Claude Mythos 5
Claude Fable 5 (`claude-fable-5`) is Anthropic's most capable widely released model. Claude Mythos 5 (`claude-mythos-5`) joins the invitation-only Claude Mythos Preview (`claude-mythos-preview`) within [Project Glasswing](https://anthropic.com/glasswing). See [Introducing Claude Fable 5 and Claude Mythos 5](/docs/en/about-claude/models/introducing-claude-fable-5-and-claude-mythos-5) for launch details and API changes.
| Feature | Claude Fable 5 | Claude Mythos 5 |
|:--------|:-------------|:-------------|
| **Description** | Anthropic's most capable widely released model, for the most demanding reasoning and long-horizon agentic work | Available through Project Glasswing. Successor to Claude Mythos Preview. |
| **Claude API ID** | `claude-fable-5` | `claude-mythos-5` |
| **AWS Bedrock ID** | anthropic.claude-fable-5 | Limited availability |
| **Vertex AI ID** | claude-fable-5 | Limited availability |
| **[Extended thinking](/docs/en/build-with-claude/extended-thinking)** | No | No |
| **[Adaptive thinking](/docs/en/build-with-claude/adaptive-thinking)** | Yes (always on) | Yes (always on) |
| **Context window** | <Tooltip tooltipContent="Claude Fable 5 and Claude Mythos 5 use the tokenizer introduced with Claude Opus 4.7. Compared to models before Claude Opus 4.7, the same text produces roughly 30% more tokens.">1M tokens</Tooltip> | <Tooltip tooltipContent="Claude Fable 5 and Claude Mythos 5 use the tokenizer introduced with Claude Opus 4.7. Compared to models before Claude Opus 4.7, the same text produces roughly 30% more tokens.">1M tokens</Tooltip> |
| **Max output** | 128k tokens | 128k tokens |
| **Pricing** | $10 / $50 per MTok (input / output) | $10 / $50 per MTok (input / output) |
Claude Fable 5 is generally available on the Claude API, Claude Platform on AWS, Amazon Bedrock, Vertex AI, and Microsoft Foundry beginning June 9, 2026. Claude Mythos 5 is not generally available: it is offered in limited availability to approved customers in [Project Glasswing](https://anthropic.com/glasswing), beginning the same day. For access, contact your Anthropic, AWS, or Google Cloud account team.
### Latest models comparison
| Feature | Claude Opus 4.8 | Claude Sonnet 4.6 | Claude Haiku 4.5 |
|:--------|:-------------|:------------------|:-----------------|
| **Description** | Anthropic's most capable model for complex reasoning and agentic coding | The best combination of speed and intelligence | The fastest model with near-frontier intelligence |
| **Description** | Anthropic's most capable Opus-tier model for complex reasoning and agentic coding | The best combination of speed and intelligence | The fastest model with near-frontier intelligence |
| **Claude API ID** | claude-opus-4-8 | claude-sonnet-4-6 | claude-haiku-4-5-20251001 |
| **Claude API alias** | claude-opus-4-8 | claude-sonnet-4-6 | claude-haiku-4-5 |
| **AWS Bedrock ID** | anthropic.claude-opus-4-8<sup>3</sup> | anthropic.claude-sonnet-4-6 | anthropic.claude-haiku-4-5-20251001-v1:0 |
@@ -41,7 +41,7 @@ Proxy and firewall requirements in [Network configuration](/en/network-config) a
## Decide how settings reach devices
Managed settings define policy that takes precedence over local developer configuration. Claude Code looks for them in four places and uses the first one it finds on a given device.
Managed settings define policy that takes precedence over local developer configuration. Claude Code checks the four sources below in priority order and applies the first one that returns a non-empty configuration.
| Mechanism | Delivery | Priority | Platforms |
| :- | :- | :- | :- |
@@ -177,7 +177,7 @@ The permission mode option (`permission_mode` in Python, `permissionMode` in Typ
| :- | :- |
| `"default"` | Tools not covered by allow rules trigger your approval callback; no callback means deny |
| `"acceptEdits"` | Auto-approves file edits and common filesystem commands (`mkdir`, `touch`, `mv`, `cp`, etc.); other Bash commands follow default rules |
| `"plan"` | Read-only tools run; Claude explores and produces a plan without editing your source files |
| `"plan"` | Claude explores and plans without editing your source files; file edits are never auto-approved and prompt through your `canUseTool` callback |
| `"dontAsk"` | Never prompts. Tools pre-approved by [permission rules](/en/settings#permission-settings) run, everything else is denied |
| `"auto"` (TypeScript only) | Uses a model classifier to approve or deny each tool call. See [Auto mode](/en/permission-modes#eliminate-prompts-with-auto-mode) for availability and behavior |
| `"bypassPermissions"` | Runs all allowed tools without asking, unless an explicit [`ask` rule](/en/settings#permission-settings) matches; see [How permissions are evaluated](/en/agent-sdk/permissions#how-permissions-are-evaluated) for where ask rules sit in the precedence order. Cannot be used when running as root on Unix. Use only in isolated environments where the agent's actions cannot affect systems you care about |
@@ -21,7 +21,7 @@ Check `deny` rules (from `disallowed_tools` and [settings.json](/en/settings#per
Check `ask` rules from [settings.json](/en/settings#permission-settings). If an ask rule matches, the call falls through to your [`canUseTool` callback](/en/agent-sdk/user-input) for confirmation, even in `bypassPermissions` mode. In `dontAsk` mode a matching ask rule is denied instead, because that mode never prompts.
Apply the active [permission mode](#permission-modes). `bypassPermissions` approves everything that reaches this step. `acceptEdits` approves file operations. Other modes fall through.
Apply the active [permission mode](#permission-modes). `bypassPermissions` approves everything that reaches this step. `acceptEdits` approves file operations. `plan` routes file-edit and shell-write tools to your `canUseTool` callback regardless of allow rules, so write operations cannot be auto-approved while planning. Other modes fall through.
Check `allow` rules (from `allowed_tools` and settings.json). If a rule matches, the tool is approved.
@@ -72,7 +72,7 @@ The SDK supports these permission modes:
| `dontAsk` | Deny instead of prompting | Anything not pre-approved by `allowed_tools` or rules is denied; `canUseTool` is never called |
| `acceptEdits` | Auto-accept file edits | File edits and [filesystem operations](#accept-edits-mode-acceptedits) (`mkdir`, `rm`, `mv`, etc.) are automatically approved |
| `bypassPermissions` | Bypass permission checks | Tools run without permission prompts, unless an explicit [`ask` rule](#how-permissions-are-evaluated) matches (use with caution) |
| `plan` | Planning mode | Read-only tools run; Claude analyzes and plans without editing your source files |
| `plan` | Planning mode | Claude explores and plans without editing your source files; file edits are never auto-approved and prompt through your `canUseTool` callback |
| `auto` (TypeScript only) | Model-classified approvals | A model classifier approves or denies each tool call. See [Auto mode](/en/permission-modes#eliminate-prompts-with-auto-mode) for availability |
**Subagent inheritance:** When the parent uses `bypassPermissions`, `acceptEdits`, or `auto`, all subagents inherit that mode and it cannot be overridden per subagent. Subagents may have different system prompts and less constrained behavior than your main agent, so inheriting `bypassPermissions` grants them full, autonomous system access. An explicit [`ask` rule](#how-permissions-are-evaluated) still forces a prompt.
@@ -200,7 +200,7 @@ Use with extreme caution. Claude has full system access in this mode. Only use i
#### Plan mode (`plan`)
Restricts Claude to read-only tools. Claude can read files and run read-only shell commands to explore the codebase but does not edit your source files. Claude may use `AskUserQuestion` to clarify requirements before finalizing the plan. See [Handle approvals and user input](/en/agent-sdk/user-input#handle-clarifying-questions) for handling these prompts.
Claude explores the codebase and produces a plan without editing your source files. Read-only tools run as in default mode. File edits are never auto-approved in plan mode, even when an allow rule matches. They prompt through your `canUseTool` callback instead. Claude may use `AskUserQuestion` to clarify requirements before finalizing the plan. See [Handle approvals and user input](/en/agent-sdk/user-input#handle-clarifying-questions) for handling these prompts.
**Use when:** you want Claude to propose changes without executing them, such as during code review or when you need to approve changes before they're made.
@@ -1044,7 +1044,7 @@ Permission modes for controlling tool execution.
PermissionMode = Literal[
"default", # Standard permission behavior
"acceptEdits", # Auto-accept file edits
"plan", # Planning mode - read-only tools only
"plan", # Planning mode - explore without editing
"dontAsk", # Deny anything not pre-approved instead of prompting
"bypassPermissions", # Bypass permission checks; explicit ask rules still prompt (use with caution)
]
@@ -24,7 +24,7 @@ Tool search adds one extra round-trip the first time Claude discovers a tool (th
For details on the underlying API mechanism, see [Tool search in the API](https://platform.claude.com/docs/en/agents-and-tools/tool-use/tool-search-tool).
Tool search requires Claude Sonnet 4 or later, or Claude Opus 4 or later. Haiku models do not support tool search.
Tool search is supported on every Claude model except Haiku.
## Configure tool search
@@ -113,7 +113,7 @@ You can search for tools to interact with Slack, GitHub, and Jira.
- **Maximum tools:** 10,000 tools in your catalog
- **Search results:** Returns 3-5 most relevant tools per search
- **Model support:** Claude Sonnet 4 and later, Claude Opus 4 and later (no Haiku)
- **Model support:** every Claude model except Haiku
## Related documentation
@@ -770,7 +770,7 @@ type PermissionMode =
| "default" // Standard permission behavior
| "acceptEdits" // Auto-accept file edits
| "bypassPermissions" // Bypass permission checks; explicit ask rules still prompt
| "plan" // Planning mode - read-only tools only
| "plan" // Planning mode - explore without editing
| "dontAsk" // Don't prompt for permissions, deny if not pre-approved
| "auto"; // Use a model classifier to approve or deny each tool call
```
@@ -1721,7 +1721,7 @@ type AgentInput = {
description: string;
prompt: string;
subagent_type: string;
model?: "sonnet" | "opus" | "haiku";
model?: "sonnet" | "opus" | "haiku" | "fable";
resume?: string;
run_in_background?: boolean;
max_turns?: number;
@@ -300,6 +300,7 @@ Environments run behind an HTTP/HTTPS network proxy for security and abuse preve
- Protection against malicious requests
- Rate limiting and abuse prevention
- Content filtering for enhanced security
- A DNS-level audit trail of requested hostnames
### Default allowed domains
@@ -86,7 +86,7 @@ Customize Claude Code's behavior with these command-line flags. `claude --help`
| `--max-budget-usd` | Maximum dollar amount to spend on API calls before stopping (print mode only) | `claude -p --max-budget-usd 5.00 "query"` |
| `--max-turns` | Limit the number of agentic turns (print mode only). Exits with an error when the limit is reached. No limit by default | `claude -p --max-turns 3 "query"` |
| `--mcp-config` | Load MCP servers from JSON files or strings (space-separated) | `claude --mcp-config ./mcp.json` |
| `--model` | Sets the model for the current session with an alias for the latest model (`sonnet` or `opus`) or a model's full name. Overrides the [`model`](/en/settings#available-settings) setting and [`ANTHROPIC_MODEL`](/en/model-config#environment-variables) | `claude --model claude-sonnet-4-6` |
| `--model` | Sets the model for the current session with an alias for the latest model (`sonnet`, `opus`, `haiku`, or `fable`) or a model's full name. Overrides the [`model`](/en/settings#available-settings) setting and [`ANTHROPIC_MODEL`](/en/model-config#environment-variables) | `claude --model claude-sonnet-4-6` |
| `--name`, `-n` | Set a display name for the session, shown in `/resume` and the terminal title. You can resume a named session with `claude --resume <name>`. [`/rename`](/en/commands) changes the name mid-session and also shows it on the prompt bar | `claude -n "my-feature-work"` |
| `--no-chrome` | Disable [Chrome browser integration](/en/chrome) for this session | `claude --no-chrome` |
| `--no-session-persistence` | Disable session persistence so sessions are not saved to disk and cannot be resumed. Print mode only. The [`CLAUDE_CODE_SKIP_PROMPT_HISTORY`](/en/env-vars) environment variable does the same in any mode | `claude -p --no-session-persistence "query"` |
@@ -102,7 +102,7 @@ Customize Claude Code's behavior with these command-line flags. `claude --help`
| `--remote-control-session-name-prefix <prefix>` | Prefix for auto-generated [Remote Control](/en/remote-control) session names when no explicit name is set. Defaults to your machine's hostname, producing names like `myhost-graceful-unicorn`. Set `CLAUDE_REMOTE_CONTROL_SESSION_NAME_PREFIX` for the same effect | `claude remote-control --remote-control-session-name-prefix dev-box` |
| `--replay-user-messages` | Re-emit user messages from stdin back on stdout for acknowledgment. Requires `--input-format stream-json` and `--output-format stream-json` | `claude -p --input-format stream-json --output-format stream-json --verbose --replay-user-messages` |
| `--resume`, `-r` | Resume a specific session by ID or name, or show an interactive picker to choose a session. Includes sessions that added this directory with `/add-dir`. As of v2.1.144, [background sessions](/en/agent-view) appear in the picker marked with `bg` | `claude --resume auth-refactor` |
| `--safe-mode` | Start with all customizations disabled to troubleshoot a broken configuration: CLAUDE.md, skills, plugins, hooks, MCP servers, custom commands and agents, output styles, workflows, custom themes, custom keybindings, status line and file-suggestion commands, LSP servers, and auto-memory do not load. Authentication, model selection, built-in tools, and permissions work normally, which differs from [`--bare`](/en/headless#start-faster-with-bare-mode). Managed settings policy still applies, including policy-configured hooks, status line, and file-suggestion commands; managed plugins, managed skills, managed CLAUDE.md, and policy-configured MCP servers do not. Sets [`CLAUDE_CODE_SAFE_MODE`](/en/env-vars) | `claude --safe-mode` |
| `--safe-mode` | Start with all customizations disabled to troubleshoot a broken configuration: CLAUDE.md, skills, plugins, hooks, MCP servers, custom commands and agents, output styles, workflows, custom themes, custom keybindings, status line and file-suggestion commands, LSP servers, and auto-memory do not load. Authentication, model selection, built-in tools, and permissions work normally, which differs from [`--bare`](/en/headless#start-faster-with-bare-mode). Managed settings policy still applies, including policy-configured hooks, status line, and file-suggestion commands; managed plugins, managed skills, managed CLAUDE.md, and policy-configured MCP servers do not. Useful for checking whether a customization is what triggers [automatic fallback from Fable 5](/en/model-config#automatic-model-fallback). Sets [`CLAUDE_CODE_SAFE_MODE`](/en/env-vars) | `claude --safe-mode` |
| `--session-id` | Use a specific session ID for the conversation (must be a valid UUID) | `claude --session-id "550e8400-e29b-41d4-a716-446655440000"` |
| `--setting-sources` | Comma-separated list of setting sources to load (`user`, `project`, `local`) | `claude --setting-sources user,project` |
| `--settings` | Path to a settings JSON file or an inline JSON string. Values you set here override the same keys in your `settings.json` files for this session. Keys you omit keep their file-based values. See [settings precedence](/en/settings#settings-precedence) | `claude --settings ./settings.json` |
@@ -200,7 +200,10 @@ Claude Code runs on the same models as the Claude app, and you can switch
mid-session. *Sonnet* is the workhorse default for everyday feature work,
bugs, tests, and reviews. Reach for *Opus* on large refactors, gnarly
debugging, or anything high-stakes. Drop to *Haiku* for quick questions,
formatting, and mechanical edits where speed wins.
formatting, and mechanical edits where speed wins. *Fable 5* is the most
capable model for your hardest, longest-running tasks; it is not the
default, so select it with `/model fable`, and note that cybersecurity and
biology content falls back to Opus automatically.
*Try it now:* type `/model` and pick Sonnet if you haven't already. It is
the right default for most tasks.
@@ -210,6 +213,7 @@ the right default for most tasks.
| Model | Best for |
| - | - |
| Fable 5 | The hardest, longest-running tasks. Opt-in only: select it with `/model fable`. Cybersecurity or biology content [falls back to Opus](/en/model-config#automatic-model-fallback) |
| Opus | Large-scale refactors, complex debugging, architecture decisions, high-stakes changes |
| Sonnet | Everyday feature work, bug fixes, tests, documentation, code review. Recommended default. |
| Haiku | Quick questions, formatting, mechanical edits, rapid iteration |
@@ -46,7 +46,7 @@ You can also act before the automatic pass runs:
- **Clear between tasks**: run `/clear` when switching to unrelated work. Old conversation crowds out the files you need next and costs tokens on every message.
- **Delegate large reads**: send research to a [subagent](/en/sub-agents) so the file contents stay in its context window, not yours.
If you need a larger window rather than a smaller conversation, Opus 4.6 and later, and Sonnet 4.6, support a 1 million token context window. See [Extended context](/en/model-config#extended-context) for availability by plan and how to select a `[1m]` model variant. Compaction works the same way at the larger limit.
If you need a larger window rather than a smaller conversation, Fable 5, Opus 4.6 and later, and Sonnet 4.6 support a 1 million token context window. See [Extended context](/en/model-config#extended-context) for availability by plan and how to select a `[1m]` model variant. Compaction works the same way at the larger limit.
## Check your own session
@@ -159,7 +159,7 @@ Your [CLAUDE.md](/en/memory) file is loaded into context at session start. If it
### Adjust extended thinking
Extended thinking is enabled by default because it significantly improves performance on complex planning and reasoning tasks. Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model. For simpler tasks where deep reasoning isn't needed, you can reduce costs by lowering the [effort level](/en/model-config#adjust-effort-level) with `/effort` or in `/model`, disabling thinking in `/config`, or, on models with a [fixed thinking budget](/en/model-config#adaptive-reasoning-and-fixed-thinking-budgets), lowering the budget with `MAX_THINKING_TOKENS=8000`. Adaptive-reasoning models ignore nonzero budgets, so use effort levels there instead.
Extended thinking is enabled by default because it significantly improves performance on complex planning and reasoning tasks. Thinking tokens are billed as output tokens, and the default budget can be tens of thousands of tokens per request depending on the model. For simpler tasks where deep reasoning isn't needed, you can reduce costs by lowering the [effort level](/en/model-config#adjust-effort-level) with `/effort` or in `/model`, disabling thinking in `/config`, or, on models with a [fixed thinking budget](/en/model-config#adaptive-reasoning-and-fixed-thinking-budgets), lowering the budget with `MAX_THINKING_TOKENS=8000`. Adaptive-reasoning models ignore nonzero budgets, so use effort levels there instead. Disabling thinking is not available on Fable 5, which always uses extended thinking.
### Delegate verbose operations to subagents
@@ -48,7 +48,7 @@ You can also select:
- **Remote**: Run sessions on Anthropic's cloud infrastructure that continue even if you close the app. Cloud sessions use the same infrastructure as [Claude Code on the web](/en/claude-code-on-the-web).
- **SSH**: Connect to a remote machine over SSH, such as your own servers, cloud VMs, or dev containers. Desktop installs Claude Code on the remote machine automatically the first time you connect.
Select a model from the dropdown next to the send button. See [models](/en/model-config#available-models) for a comparison of Opus, Sonnet, and Haiku. You can change the model later from the same dropdown.
Select a model from the dropdown next to the send button. See [models](/en/model-config#available-models) for a comparison of the available models. You can change the model later from the same dropdown.
Type what you want Claude to do:
@@ -485,7 +485,7 @@ The desktop app does not always inherit your full shell environment. On macOS, w
To set environment variables for local sessions and dev servers on any platform, open the environment dropdown in the prompt box, hover over **Local**, and click the gear icon to open the local environment editor. Variables you save here are stored encrypted on your machine and apply to every local session and preview server you start. You can also add variables to the `env` key in your `~/.claude/settings.json` file, though these reach Claude sessions only and not dev servers. See [environment variables](/en/env-vars) for the full list of supported variables.
[Extended thinking](/en/model-config#extended-thinking) is enabled by default, which improves performance on complex reasoning tasks but uses additional tokens. To disable thinking, set `MAX_THINKING_TOKENS` to `0` in the local environment editor. On [third-party providers](/en/third-party-integrations), `0` omits the `thinking` parameter instead, and adaptive-reasoning models may still think. On models with [adaptive reasoning](/en/model-config#adjust-effort-level), any other `MAX_THINKING_TOKENS` value is ignored because adaptive reasoning controls thinking depth instead. On Opus 4.6 and Sonnet 4.6, set `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` to `1` to use a fixed thinking budget; Opus 4.7 and later always use adaptive reasoning and have no fixed-budget mode.
[Extended thinking](/en/model-config#extended-thinking) is enabled by default, which improves performance on complex reasoning tasks but uses additional tokens. To disable thinking, set `MAX_THINKING_TOKENS` to `0` in the local environment editor; this has no effect on Fable 5, which always uses extended thinking. On [third-party providers](/en/third-party-integrations), `0` omits the `thinking` parameter instead, and adaptive-reasoning models may still think. On models with [adaptive reasoning](/en/model-config#adjust-effort-level), any other `MAX_THINKING_TOKENS` value is ignored because adaptive reasoning controls thinking depth instead. On Opus 4.6 and Sonnet 4.6, set `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` to `1` to use a fixed thinking budget; Opus 4.7 and later always use adaptive reasoning and have no fixed-budget mode.
### Cloud sessions
@@ -581,7 +581,7 @@ A managed settings file deployed to disk on each machine applies to Desktop sess
IT teams can manage the desktop app through MDM on macOS or group policy on Windows. Available policies include enabling or disabling the Claude Code feature, controlling auto-updates, and setting a custom deployment URL.
- **macOS**: configure via `com.anthropic.Claude` preference domain using tools like Jamf or Kandji
- **macOS**: configure via `com.anthropic.claudefordesktop` preference domain using tools like Jamf or Kandji
- **Windows**: configure via registry at `SOFTWARE\Policies\Claude`
### Authentication and SSO
@@ -636,7 +636,7 @@ Desktop and CLI read the same configuration files, so your setup carries over:
- **[MCP servers](/en/mcp)** configured in `~/.claude.json` or `.mcp.json` work in both
- **[Hooks](/en/hooks)** and **[skills](/en/skills)** defined in settings apply to both
- **[Settings](/en/settings)** in `~/.claude.json` and `~/.claude/settings.json` are shared. Permission rules, allowed tools, and other settings in `settings.json` apply to Desktop sessions.
- **Models**: Sonnet, Opus, and Haiku are available in both. In Desktop, select the model from the dropdown next to the send button. You can change the model mid-session from the same dropdown.
- **Models**: the same [models](/en/model-config#available-models) are available in both. In Desktop, select the model from the dropdown next to the send button. You can change the model mid-session from the same dropdown.
**MCP servers from the Claude Desktop chat app**: the Desktop app loads MCP servers from `claude_desktop_config.json` into Code tab sessions, alongside servers from `~/.claude.json` and `.mcp.json`. A server defined in `claude_desktop_config.json` is available in both the Desktop chat surface and the Code tab.
@@ -95,6 +95,10 @@ Claude Code reads environment variables at startup, so changes take effect the n
| `ANTHROPIC_CUSTOM_MODEL_OPTION_DESCRIPTION` | Display description for the custom model entry in the `/model` picker. Defaults to `Custom model (<model-id>)` when not set |
| `ANTHROPIC_CUSTOM_MODEL_OPTION_NAME` | Display name for the custom model entry in the `/model` picker. Defaults to the model ID when not set |
| `ANTHROPIC_CUSTOM_MODEL_OPTION_SUPPORTED_CAPABILITIES` | See [Model configuration](/en/model-config#customize-pinned-model-display-and-capabilities) |
| `ANTHROPIC_DEFAULT_FABLE_MODEL` | See [Model configuration](/en/model-config#environment-variables) |
| `ANTHROPIC_DEFAULT_FABLE_MODEL_DESCRIPTION` | See [Model configuration](/en/model-config#customize-pinned-model-display-and-capabilities) |
| `ANTHROPIC_DEFAULT_FABLE_MODEL_NAME` | See [Model configuration](/en/model-config#customize-pinned-model-display-and-capabilities) |
| `ANTHROPIC_DEFAULT_FABLE_MODEL_SUPPORTED_CAPABILITIES` | See [Model configuration](/en/model-config#customize-pinned-model-display-and-capabilities) |
| `ANTHROPIC_DEFAULT_HAIKU_MODEL` | See [Model configuration](/en/model-config#environment-variables) |
| `ANTHROPIC_DEFAULT_HAIKU_MODEL_DESCRIPTION` | See [Model configuration](/en/model-config#customize-pinned-model-display-and-capabilities) |
| `ANTHROPIC_DEFAULT_HAIKU_MODEL_NAME` | See [Model configuration](/en/model-config#customize-pinned-model-display-and-capabilities) |
@@ -144,7 +148,7 @@ Claude Code reads environment variables at startup, so changes take effect the n
| `CLAUDE_CODE_DEBUG_LOGS_DIR` | Override the debug log file path. Despite the name, this is a file path, not a directory. Requires debug mode to be enabled separately via `--debug`, `/debug`, or the `DEBUG` environment variable: setting this variable alone does not enable logging. The [`--debug-file`](/en/cli-reference#cli-flags) flag does both at once. Defaults to `~/.claude/debug/<session-id>.txt` |
| `CLAUDE_CODE_DEBUG_LOG_LEVEL` | Minimum log level written to the debug log file. Values: `verbose`, `debug` (default), `info`, `warn`, `error`. Set to `verbose` to include high-volume diagnostics like full status line command output, or raise to `error` to reduce noise |
| `CLAUDE_CODE_DISABLE_1M_CONTEXT` | Set to `1` to disable [1M context window](/en/model-config#extended-context) support. When set, 1M model variants are unavailable in the model picker. Useful for enterprise environments with compliance requirements |
| `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` | Set to `1` to disable [adaptive reasoning](/en/model-config#adjust-effort-level) on Opus 4.6 and Sonnet 4.6 and fall back to the fixed thinking budget controlled by `MAX_THINKING_TOKENS`. From v2.1.111, has no effect on Opus 4.7 and later, which always use adaptive reasoning |
| `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` | Set to `1` to disable [adaptive reasoning](/en/model-config#adjust-effort-level) on Opus 4.6 and Sonnet 4.6 and fall back to the fixed thinking budget controlled by `MAX_THINKING_TOKENS`. From v2.1.111, has no effect on Fable 5, or on Opus 4.7 and later, which always use adaptive reasoning |
| `CLAUDE_CODE_DISABLE_AGENT_VIEW` | Set to `1` to turn off [background agents and agent view](/en/agent-view): `claude agents`, `--bg`, `/background`, and the on-demand supervisor. Equivalent to the [`disableAgentView`](/en/settings#available-settings) setting |
| `CLAUDE_CODE_DISABLE_ALTERNATE_SCREEN` | Set to `1` to disable [fullscreen rendering](/en/fullscreen) and use the classic main-screen renderer. The conversation stays in your terminal's native scrollback so `Cmd+f` and tmux copy mode work as usual. Takes precedence over `CLAUDE_CODE_NO_FLICKER` and the [`tui`](/en/settings#available-settings) setting. You can also switch with `/tui default`. Does not apply to background sessions opened from [agent view](/en/agent-view), which always use fullscreen rendering |
| `CLAUDE_CODE_DISABLE_ATTACHMENTS` | Set to `1` to disable attachment processing. File mentions with `@` syntax are sent as plain text instead of being expanded into file content |
@@ -165,7 +169,7 @@ Claude Code reads environment variables at startup, so changes take effect the n
| `CLAUDE_CODE_DISABLE_OFFICIAL_MARKETPLACE_AUTOINSTALL` | Set to `1` to skip automatic addition of the official plugin marketplace on first run |
| `CLAUDE_CODE_DISABLE_POLICY_SKILLS` | Set to `1` to skip loading skills from the system-wide managed skills directory. Useful for container or CI sessions that should not load operator-provisioned skills |
| `CLAUDE_CODE_DISABLE_TERMINAL_TITLE` | Set to `1` to disable automatic terminal title updates based on conversation context. In Agent SDK and `claude -p` sessions, this also skips the background Haiku request that generates the session title |
| `CLAUDE_CODE_DISABLE_THINKING` | Set to `1` to omit the `thinking` parameter from API requests entirely. This is a compatibility option for proxies and gateways that reject the parameter. The variable's behavior is unchanged from earlier versions; on models that think by default, omitting the parameter means the model may still think. To explicitly disable [extended thinking](https://platform.claude.com/docs/en/build-with-claude/extended-thinking) on the Anthropic API, use `MAX_THINKING_TOKENS=0` instead. On [third-party providers](/en/third-party-integrations), `0` likewise omits the parameter, so the two variables behave the same there |
| `CLAUDE_CODE_DISABLE_THINKING` | Set to `1` to omit the `thinking` parameter from API requests entirely. This is a compatibility option for proxies and gateways that reject the parameter. The variable's behavior is unchanged from earlier versions; on models that think by default, omitting the parameter means the model may still think. To explicitly disable [extended thinking](https://platform.claude.com/docs/en/build-with-claude/extended-thinking) on the Anthropic API, use `MAX_THINKING_TOKENS=0` instead, which is also ineffective on Fable 5 since it cannot have thinking turned off. On [third-party providers](/en/third-party-integrations), `0` likewise omits the parameter, so the two variables behave the same there |
| `CLAUDE_CODE_DISABLE_VIRTUAL_SCROLL` | Set to `1` to disable virtual scrolling in [fullscreen rendering](/en/fullscreen) and render every message in the transcript. Use this if scrolling in fullscreen mode shows blank regions where messages should appear |
| `CLAUDE_CODE_DISABLE_WORKFLOWS` | Set to `1` to disable [workflows](/en/workflows#turn-workflows-off). Equivalent to the [`disableWorkflows`](/en/settings#available-settings) setting |
| `CLAUDE_CODE_EFFORT_LEVEL` | Set the effort level for supported models. Values: `low`, `medium`, `high`, `xhigh`, `max`, or `auto` to use the model default. Available levels depend on the model. Takes precedence over `/effort` and the `effortLevel` setting. See [Adjust effort level](/en/model-config#adjust-effort-level) |
@@ -303,7 +307,7 @@ Claude Code reads environment variables at startup, so changes take effect the n
| `IS_DEMO` | Set to `1` to enable demo mode: hides your email and organization name from the header and `/status` output, and skips onboarding. Useful when streaming or recording a session |
| `MAX_MCP_OUTPUT_TOKENS` | Maximum number of tokens allowed in MCP tool responses. Claude Code displays a warning when output exceeds 10,000 tokens. Tools that declare [`anthropic/maxResultSizeChars`](/en/mcp#raise-the-limit-for-a-specific-tool) use that character limit for text content instead, but image content from those tools is still subject to this variable (default: 25000) |
| `MAX_STRUCTURED_OUTPUT_RETRIES` | Number of times to retry when the model's response fails validation against the [`--json-schema`](/en/cli-reference#cli-flags) in non-interactive mode (the `-p` flag). Defaults to 5 |
| `MAX_THINKING_TOKENS` | Override the [extended thinking](https://platform.claude.com/docs/en/build-with-claude/extended-thinking) token budget. The ceiling is the model's [max output tokens](https://platform.claude.com/docs/en/about-claude/models/overview#latest-models-comparison) minus one. Set to `0` to disable thinking on the Anthropic API. On [third-party providers](/en/third-party-integrations), `0` omits the `thinking` parameter instead, and models with [adaptive reasoning](/en/model-config#adjust-effort-level) may still think. For nonzero values on adaptive reasoning models, the budget is ignored unless adaptive reasoning is disabled via `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` |
| `MAX_THINKING_TOKENS` | Override the [extended thinking](https://platform.claude.com/docs/en/build-with-claude/extended-thinking) token budget. The ceiling is the model's [max output tokens](https://platform.claude.com/docs/en/about-claude/models/overview#latest-models-comparison) minus one. Set to `0` to disable thinking on the Anthropic API except on Fable 5, which cannot have thinking turned off. On [third-party providers](/en/third-party-integrations), `0` omits the `thinking` parameter instead, and models with [adaptive reasoning](/en/model-config#adjust-effort-level) may still think. For nonzero values on adaptive reasoning models, the budget is ignored unless adaptive reasoning is disabled via `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` |
| `MCP_CLIENT_SECRET` | OAuth client secret for MCP servers that require [pre-configured credentials](/en/mcp#use-pre-configured-oauth-credentials). Avoids the interactive prompt when adding a server with `--client-secret` |
| `MCP_CONNECTION_NONBLOCKING` | Controls whether startup waits for MCP servers to connect before the first query. As of Claude Code v2.1.142, MCP startup is non-blocking by default: servers connect in the background and their tools become available as they finish. Set to `0` to restore the blocking 5-second connection wait. Servers configured with [`alwaysLoad: true`](/en/mcp#exempt-a-server-from-deferral) still block startup regardless, since their tools must be present when the first prompt is built |
| `MCP_CONNECT_TIMEOUT_MS` | How long blocking MCP startup waits, in milliseconds, for the connection batch before snapshotting the tool list (default: 5000). Applies when `MCP_CONNECTION_NONBLOCKING=0` or for servers marked [`alwaysLoad: true`](/en/mcp#exempt-a-server-from-deferral). Servers still pending at the deadline keep connecting in the background but won't appear until the next query. Distinct from `MCP_TIMEOUT`, which bounds an individual server's connect attempt |
@@ -653,7 +653,13 @@ The check evaluates the full conversation, not only your latest prompt, so sendi
## Responses seem lower quality than usual
If Claude's answers seem less capable than you expect but no error is shown, the cause is usually conversation state rather than the model itself. Claude Code does not silently change model versions. It can switch to a fallback model in specific cases such as an Opus quota being reached or a Bedrock or Vertex AI region lacking your model; the Model selection check below catches both, and [Model configuration](/en/model-config) explains when fallback applies.
If Claude's answers seem less capable than you expect but no error is shown, the cause is usually conversation state rather than the model itself. Claude Code does not silently change model versions. It can switch to a fallback model in three specific cases:
- A configured [`--fallback-model`](/en/cli-reference#cli-flags) takes over after an availability error, for that turn only, with a notice in the transcript
- A Bedrock or Vertex AI startup check finds your default model unavailable
- [Automatic model fallback](/en/model-config#automatic-model-fallback) on Fable 5 moves the session to the default Opus model and shows a notice in the transcript
The Model selection check below catches the second and third cases; the first appears as a transcript notice rather than a `/model` change. [Model configuration](/en/model-config) explains when each fallback applies.
Check these first:
@@ -11,8 +11,6 @@ Claude Code GitHub Actions brings AI-powered automation to your GitHub workflow.
Claude Code GitHub Actions is built on top of the [Claude Agent SDK](/en/agent-sdk/overview), which enables programmatic integration of Claude Code into your applications. You can use the SDK to build custom automation workflows beyond GitHub Actions.
**Claude Opus 4.8 is now available.** Claude Code GitHub Actions default to Sonnet. To use Opus 4.8, configure the [model parameter](#breaking-changes-reference) to use `claude-opus-4-8`.
## Why use Claude Code GitHub Actions?
- **Instant PR creation**: Describe what you need, and Claude creates a complete PR with all necessary changes
@@ -119,7 +119,7 @@ Learn more: [Sessions from Dispatch](/en/desktop#sessions-from-dispatch)
### Effort level
A setting that controls how much of the adaptive-reasoning thinking budget Claude uses on each turn. Higher effort means more thinking tokens and deeper reasoning; lower effort is faster and cheaper. Effort is supported on Opus 4.6 and later, and on Sonnet 4.6.
A setting that controls how much of the adaptive-reasoning thinking budget Claude uses on each turn. Higher effort means more thinking tokens and deeper reasoning; lower effort is faster and cheaper. Effort is supported on Fable 5, on Opus 4.6 and later, and on Sonnet 4.6.
Learn more: [Adjust effort level](/en/model-config#adjust-effort-level)
@@ -213,7 +213,7 @@ Learn more: [The `.claude` directory](/en/claude-directory)
### Prompt injection
Hostile instructions embedded in a file, web page, or tool result that attempt to redirect Claude toward actions you never asked for. Claude Code's defenses include the permission system, command blocklists, and trust verification. [Auto mode](#auto-mode) adds a server-side probe that scans tool results for suspicious content and a classifier that never sees tool results, so injected text cannot influence its approval decisions.
Hostile instructions embedded in a file, web page, or tool result that attempt to redirect Claude toward actions you never asked for. Claude Code's defenses include the permission system, command injection detection, and trust verification. [Auto mode](#auto-mode) adds a server-side probe that scans tool results for suspicious content and a classifier that never sees tool results, so injected text cannot influence its approval decisions.
Learn more: [Protect against prompt injection](/en/security#protect-against-prompt-injection)
@@ -208,7 +208,7 @@ claude -p "Look at my staged changes and create an appropriate commit" \
The `--allowedTools` flag uses [permission rule syntax](/en/settings#permission-rule-syntax). The trailing ` *` enables prefix matching, so `Bash(git diff *)` allows any command starting with `git diff`. The space before `*` is important: without it, `Bash(git diff*)` would also match `git diff-index`.
User-invoked [skills](/en/skills) like `/code-review` and [built-in commands](/en/commands) are only available in interactive mode. In `-p` mode, describe the task you want to accomplish instead.
User-invoked [skills](/en/skills) and custom commands work in `-p` mode: include `/skill-name` in the prompt string and Claude Code expands it before running. Built-in commands that open an interactive dialog, such as `/config` and `/login`, are not available in `-p` mode.
### Customize the system prompt
@@ -451,7 +451,7 @@ Each hook has a `type` that determines how it runs. Most hooks use `"type": "com
When multiple hooks match the same event, every hook's command runs to completion before Claude Code merges the results. One hook returning `deny` does not stop sibling hooks from executing. Don't rely on one hook's `deny` to suppress side effects in another hook.
After all matching hooks finish, Claude Code combines their outputs. For `PreToolUse` permission decisions, the most restrictive answer wins: `deny` overrides `ask`, which overrides `allow`. Text from `additionalContext` is kept from every hook and passed to Claude together.
After all matching hooks finish, Claude Code combines their outputs. For `PreToolUse` permission decisions, the most restrictive answer wins, in the order `deny`, `defer`, `ask`, `allow`. Text from `additionalContext` is kept from every hook and passed to Claude together.
The example below registers two `PreToolUse` hooks on `Bash`. The first appends every command to a log file and exits 0. The second runs a script that exits 2 to deny when the command contains `rm -rf`:
@@ -768,6 +768,15 @@ Not every event supports blocking or controlling behavior through JSON. The even
| SessionStart, Setup, SubagentStart | Context only | `hookSpecificOutput.additionalContext` adds context for Claude. SessionStart also accepts [`initialUserMessage`, `watchPaths`, `sessionTitle`, and `reloadSkills`](#sessionstart-decision-control). No blocking or decision control |
| WorktreeRemove, Notification, SessionEnd, PostCompact, InstructionsLoaded, StopFailure, CwdChanged, FileChanged | None | No decision control. Used for side effects like logging or cleanup |
A few events can also rewrite content rather than only allow or block it:
- `PreToolUse` — `updatedInput` directly under `hookSpecificOutput` replaces a tool's arguments before it runs ([details](#pretooluse-decision-control))
- `PermissionRequest` — `updatedInput` inside the `decision` object ([details](#permissionrequest-decision-control))
- `PostToolUse` — `updatedToolOutput` replaces the tool's result ([details](#posttooluse-decision-control))
- `UserPromptSubmit` — cannot replace the prompt; only injects `additionalContext` alongside it
For redaction or transformation use cases, intercept at `PreToolUse` for outbound tool inputs and `PostToolUse` for inbound tool results.
Here are examples of each pattern in action:
Used by `UserPromptSubmit`, `UserPromptExpansion`, `PostToolUse`, `PostToolUseFailure`, `PostToolBatch`, `Stop`, `SubagentStop`, `ConfigChange`, and `PreCompact`. The only value is `"block"`. To allow the action to proceed, omit `decision` from your JSON, or exit 0 without any JSON at all:
@@ -153,7 +153,7 @@ Press `Shift+Tab` to cycle through permission modes:
- **Default**: Claude asks before file edits and shell commands
- **Auto-accept edits**: Claude edits files and runs common filesystem commands like `mkdir` and `mv` without asking, still asks for other commands
- **Plan mode**: Claude uses read-only tools only, creating a plan you can approve before execution
- **Plan mode**: Claude explores and proposes a plan without editing your source files; permission prompts still apply as in default mode
- **Auto mode**: Claude evaluates all actions with background safety checks. Currently a research preview
You can also allow specific commands in `.claude/settings.json` so Claude doesn't ask each time. This is useful for trusted commands like `npm test` or `git status`. Settings can be scoped from organization-wide policies down to personal preferences. See [Permissions](/en/permissions) for details.
@@ -39,7 +39,7 @@ See [Terminal configuration](/en/terminal-config) for details.
| `Esc` + `Esc` | Clear input draft, or rewind | When the prompt input contains text, double `Esc` clears it and saves the draft to history so `Up` recalls it. When the input is empty, double `Esc` opens the [rewind menu](/en/checkpointing) to restore or summarize code and conversation from a previous point |
| `Shift+Tab` or `Alt+M` (some configurations) | Cycle permission modes | Cycle through `default`, `acceptEdits`, `plan`, and any modes you have enabled, such as `auto` or `bypassPermissions`. See [permission modes](/en/permission-modes). |
| `Option+P` (macOS) or `Alt+P` (Windows/Linux) | Switch model | Switch models without clearing your prompt |
| `Option+T` (macOS) or `Alt+T` (Windows/Linux) | Toggle extended thinking | Enable or disable extended thinking mode. As of v2.1.132 this shortcut works on macOS without configuring Option as Meta |
| `Option+T` (macOS) or `Alt+T` (Windows/Linux) | Toggle extended thinking | Enable or disable extended thinking mode. Has no effect on Fable 5, which always uses extended thinking. As of v2.1.132 this shortcut works on macOS without configuring Option as Meta |
| `Option+O` (macOS) or `Alt+O` (Windows/Linux) | Toggle fast mode | Enable or disable [fast mode](/en/fast-mode) |
### Text editing
@@ -882,7 +882,7 @@ Claude Code truncates tool descriptions and server instructions at 2KB each. Kee
Tool search is enabled by default: MCP tools are deferred and discovered on demand. Claude Code disables it by default on Vertex AI. It is also disabled when `ANTHROPIC_BASE_URL` points to a non-first-party host, since most proxies do not forward `tool_reference` blocks. Set `ENABLE_TOOL_SEARCH` explicitly to override either fallback.
Tool search requires a model that supports `tool_reference` blocks: Sonnet 4 and later, or Opus 4 and later. Haiku models do not support it. On Vertex AI, tool search is supported for Claude Sonnet 4.5 and later and Claude Opus 4.5 and later.
Tool search requires a model that supports `tool_reference` blocks. Haiku models do not support it. On Vertex AI, tool search is supported for Claude Sonnet 4.5 and later and Claude Opus 4.5 and later.
Control tool search behavior with the `ENABLE_TOOL_SEARCH` environment variable:
@@ -28,7 +28,8 @@ remembering exact version numbers:
| Model alias | Behavior |
| - | - |
| **`default`** | Special value that clears any model override and reverts to the recommended model for your account type. Not itself a model alias |
| **`best`** | Uses the most capable available model, currently equivalent to `opus` |
| **`best`** | Uses Fable 5 where your organization has access to it, otherwise the latest Opus model |
| **`fable`** | Uses Claude Fable 5 for your hardest and longest-running tasks |
| **`sonnet`** | Uses the latest Sonnet model for daily coding tasks |
| **`opus`** | Uses the latest Opus model for complex reasoning tasks |
| **`haiku`** | Uses the fast and efficient Haiku model for simple tasks |
@@ -42,6 +43,21 @@ Aliases point to the recommended version for your provider and update over time.
Opus 4.8 requires Claude Code v2.1.154 or later. Run `claude update` to upgrade.
### Work with Fable 5
[Claude Fable 5](https://platform.claude.com/docs/en/about-claude/models/introducing-claude-fable-5) is the most capable model in Claude Code, suited to tasks larger than a single sitting. It sustains long autonomous sessions, investigates before acting, and verifies its work more often than smaller models.
Fable 5 is not the default model. Select it with `/model fable`. Requests that its safety classifiers flag, most often in cybersecurity and biology domains, trigger [automatic model fallback](#automatic-model-fallback).
To get the most from Fable 5:
- **Describe the outcome, not the steps**: hand it the result you want and let it plan the path. To keep it working until that outcome holds, [set a goal](/en/goal).
- **Hand it ambiguous problems**: root-cause investigations, outage debugging, and architecture decisions are where the extra investigation and verification pay off.
- **Skip the verification reminders**: it verifies its own work with less prompting, so reminders to test or check are usually unnecessary.
- **Size up larger tasks**: give it work you would normally break into pieces. It holds long sessions without losing the thread.
Fable 5 requires Claude Code v2.1.170 or later. Older versions do not show Fable 5 in the model picker and cannot select it. Run `claude update` to upgrade. Fable 5 is not available under [zero data retention](/en/zero-data-retention), where the `/model` picker either omits it or shows it disabled.
### Setting your model
You can configure your model in several ways, listed in order of priority:
@@ -114,7 +130,7 @@ To fully control the model experience, combine three settings:
- **`availableModels`**: restricts which named models users can switch to
- **`model`**: sets the initial model selection when a session starts
- **`ANTHROPIC_DEFAULT_SONNET_MODEL`** / **`ANTHROPIC_DEFAULT_OPUS_MODEL`** / **`ANTHROPIC_DEFAULT_HAIKU_MODEL`**: control what the Default option and the `sonnet`, `opus`, and `haiku` aliases resolve to
- **`ANTHROPIC_DEFAULT_SONNET_MODEL`** / **`ANTHROPIC_DEFAULT_OPUS_MODEL`** / **`ANTHROPIC_DEFAULT_HAIKU_MODEL`** / **`ANTHROPIC_DEFAULT_FABLE_MODEL`**: control what the Default option and the `sonnet`, `opus`, `haiku`, and `fable` aliases resolve to
This example starts users on Sonnet 4.5, limits the picker to Sonnet and Haiku, and pins Default to resolve to Sonnet 4.5 rather than the latest release:
@@ -151,7 +167,7 @@ The behavior of `default` depends on your account type:
Enterprise pay-as-you-go means an Enterprise organization billed by usage rather than by subscription seat.
Claude Code may automatically fall back to Sonnet if you hit a usage threshold with Opus.
Fable 5 is not the default model on any account type. Sessions use Fable 5 only after you choose it, with `/model fable`, a `model` setting, or the `best` alias where Fable 5 is available. Choosing it with `/model` saves it as the selected model in your user settings, so later sessions start on Fable 5 until you change models.
### `opusplan` model setting
@@ -194,6 +210,45 @@ Two cases cause an element to be skipped:
- **Unavailable model**: a model that can't be reached, such as a retired model pinned in settings, is skipped and Claude Code continues to the next element.
- **Outside the allowlist**: an element not permitted by [`availableModels`](#restrict-model-selection) is dropped when the chain is read and never tried.
### Automatic model fallback
This section covers content-based fallback from Fable 5. For availability-based fallback when a model is overloaded or unavailable, see [Fallback model chains](#fallback-model-chains).
Fable 5 runs with safety classifiers for cybersecurity and biology content. When a classifier flags a request, Claude Code re-runs that request on the default Opus model and shows a notice in the transcript: Opus 4.8 on the Anthropic API and [LLM gateway](/en/llm-gateway) deployments, or Opus 4.7 on [Claude Platform on AWS](/en/claude-platform-on-aws).
The session then continues on that Opus model. To return to Fable 5, run `/model fable`.
#### Check what triggered fallback
Fallback can trigger on the first request of a session, before you send anything unusual, because the first request carries workspace context such as your CLAUDE.md content and git status. A repository that contains security or biology material can trip the classifier on that context alone.
To check whether customizations are the trigger, start a session with `claude --safe-mode`, which disables customizations such as CLAUDE.md, skills, MCP servers, and hooks. Git status and directory names are not customizations and are still included.
#### Ask before switching
To decide what happens each time a request is flagged, rather than switching automatically, run `/config` and turn off "switch models when a message is flagged". A flagged request then pauses the session with two options: switch to the Opus model, or edit the prompt and retry on Fable 5.
Some cases behave differently:
- If both models flag the same request, you can edit the prompt and retry, or start a new session.
- On mobile [Claude Code on the web](/en/claude-code-on-the-web) sessions, editing and retrying is not supported. Switch models, or continue the session from a desktop browser or the desktop app.
- In [non-interactive mode](/en/cli-reference#cli-flags) and SDK integrations that can't show the prompt, a flagged request ends the turn with a refusal instead.
#### Enable fallback on Bedrock, Vertex AI, and Foundry
On [Amazon Bedrock](/en/amazon-bedrock), [Google Vertex AI](/en/google-vertex-ai), and [Microsoft Foundry](/en/microsoft-foundry), model IDs are provider-specific, so automatic fallback only operates when Claude Code can identify both models involved:
- Claude Code must recognize the current model as Fable 5: the model ID contains `claude-fable-5`, matches the value of `ANTHROPIC_DEFAULT_FABLE_MODEL`, or is mapped with [`modelOverrides`](#override-model-ids-per-version).
- The fallback target must resolve to an Opus model: the value of `ANTHROPIC_DEFAULT_OPUS_MODEL` if set, otherwise an Opus 4.8 entry in the provider's model list.
If either model can't be identified, Claude Code does not switch automatically. The flagged request ends with a refusal message, and you can switch models with [`/model`](#setting-your-model) and retry. To enable automatic fallback on these providers, set `ANTHROPIC_DEFAULT_FABLE_MODEL` to your Fable 5 model ID and `ANTHROPIC_DEFAULT_OPUS_MODEL` to your Opus 4.8 model ID.
#### Security research and biology workloads
Workloads in offensive security or biology, including penetration testing, Capture the Flag (CTF) exercises, and biology-adjacent codebases, trigger fallback frequently, often on the first request. For substantive biology work, expect nearly all requests to reroute.
This is expected routing for these domains, not an account flag. If your organization needs Fable-class capability for this work, ask your Anthropic account team about trusted access programs.
### Adjust effort level
[Effort levels](https://platform.claude.com/docs/en/build-with-claude/effort) control adaptive reasoning, which lets the model decide whether and how much to think on each step based on task complexity. Lower effort is faster and cheaper for straightforward tasks, while higher effort provides deeper reasoning for complex problems.
@@ -202,14 +257,15 @@ The available effort levels depend on the model. Models not listed here do not s
| Model | Levels |
| :- | :- |
| Fable 5 | `low`, `medium`, `high`, `xhigh`, `max` |
| Opus 4.8 and Opus 4.7 | `low`, `medium`, `high`, `xhigh`, `max` |
| Opus 4.6 and Sonnet 4.6 | `low`, `medium`, `high`, `max` |
If you set a level the active model does not support, Claude Code falls back to the highest supported level at or below the one you set. For example, `xhigh` runs as `high` on Opus 4.6.
The default effort is `high` on Opus 4.8, Opus 4.6, and Sonnet 4.6, and `xhigh` on Opus 4.7.
The default effort is `high` on Fable 5, Opus 4.8, Opus 4.6, and Sonnet 4.6, and `xhigh` on Opus 4.7.
When you first run Opus 4.8 or Opus 4.7, Claude Code applies that model's default effort even if you previously set a different level for another model: `high` on Opus 4.8 and `xhigh` on Opus 4.7. Run `/effort` again to choose a different level after switching.
When you first run Fable 5, Opus 4.8, or Opus 4.7, Claude Code applies that model's default effort even if you previously set a different level for another model: `high` on Fable 5 and Opus 4.8, and `xhigh` on Opus 4.7. Run `/effort` again to choose a different level after switching.
`low`, `medium`, `high`, and `xhigh` persist across sessions. `max` provides the deepest reasoning with no constraint on token spending and applies to the current session only, except when set through the `CLAUDE_CODE_EFFORT_LEVEL` environment variable.
@@ -223,7 +279,7 @@ Each level trades token spend against capability. The default suits most coding
| :- | :- |
| `low` | Reserve for short, scoped, latency-sensitive tasks that are not intelligence-sensitive |
| `medium` | Reduces token usage for cost-sensitive work that can trade off some intelligence |
| `high` | Balances token usage and intelligence. Default on Opus 4.8, Opus 4.6, and Sonnet 4.6 |
| `high` | Balances token usage and intelligence. Default on Fable 5, Opus 4.8, Opus 4.6, and Sonnet 4.6 |
| `xhigh` | Deeper reasoning at higher token spend. Default on Opus 4.7 |
| `max` | Can improve performance on demanding tasks but may show diminishing returns and is prone to overthinking. Test before adopting broadly |
| `ultracode` | A Claude Code setting that plans a [dynamic workflow](/en/workflows) for each substantive task with `xhigh` per-message reasoning. Session-only |
@@ -253,7 +309,7 @@ The effort slider appears in `/model` when a supported model is selected. The cu
Adaptive reasoning makes thinking optional on each step, so Claude can respond faster to routine prompts and reserve deeper thinking for steps that benefit from it. If you want Claude to think more or less often than the current level produces, you can say so directly in your prompt or in `CLAUDE.md`; the model responds to that guidance within its effort setting.
Opus 4.7 and later always use adaptive reasoning. The fixed thinking budget mode and `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` do not apply to them.
Opus 4.7 and later always use adaptive reasoning, as does Fable 5. The fixed thinking budget mode and `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING` do not apply to them.
On Opus 4.6 and Sonnet 4.6, you can set `CLAUDE_CODE_DISABLE_ADAPTIVE_THINKING=1` to revert to the previous fixed thinking budget controlled by `MAX_THINKING_TOKENS`. See [environment variables](/en/env-vars).
@@ -265,15 +321,17 @@ Extended thinking is the reasoning Claude emits before responding. On models tha
| :- | :- |
| Toggle for the current session | Press `Option+T` on macOS or `Alt+T` on Windows and Linux |
| Set the global default | Run `/config` and toggle thinking mode. Saved as `alwaysThinkingEnabled` in `~/.claude/settings.json` |
| Disable regardless of effort | Set [`MAX_THINKING_TOKENS=0`](/en/env-vars). On [third-party providers](/en/third-party-integrations) this omits the `thinking` parameter instead, and adaptive-reasoning models may still think. Other values apply only with a [fixed thinking budget](#adaptive-reasoning-and-fixed-thinking-budgets) |
| Disable regardless of effort | Set [`MAX_THINKING_TOKENS=0`](/en/env-vars), which turns thinking off on the Anthropic API except on Fable 5. On [third-party providers](/en/third-party-integrations) this omits the `thinking` parameter instead, and adaptive-reasoning models may still think. Other values apply only with a [fixed thinking budget](#adaptive-reasoning-and-fixed-thinking-budgets) |
Thinking cannot be turned off on Fable 5. The session toggle, `alwaysThinkingEnabled`, and `MAX_THINKING_TOKENS=0` have no effect there, and Fable 5 decides per step how much to think based on the effort level.
Thinking output is collapsed by default. Press `Ctrl+O` to toggle verbose mode and see the reasoning as gray italic text. Interactive sessions on the Anthropic API receive redacted thinking blocks by default, so set `showThinkingSummaries: true` in [settings](/en/settings) if you want the full summaries available when you expand. You are charged for all thinking tokens generated, even when collapsed or redacted.
### Extended context
Opus 4.6 and later, and Sonnet 4.6, support a [1 million token context window](https://platform.claude.com/docs/en/build-with-claude/context-windows#1m-token-context-window) for long sessions with large codebases.
Fable 5, Opus 4.6 and later, and Sonnet 4.6 support a [1 million token context window](https://platform.claude.com/docs/en/build-with-claude/context-windows#1m-token-context-window) for long sessions with large codebases.
Availability varies by model and plan. On Max, Team, and Enterprise plans, Opus is automatically upgraded to 1M context with no additional configuration. This applies to both Team Standard and Team Premium seats. Sonnet with 1M context is not part of the automatic upgrade and requires [usage credits](https://support.claude.com/en/articles/12429409-extra-usage-for-paid-claude-plans) on every subscription plan, including Max.
Availability varies by model and plan. On Max, Team, and Enterprise plans, Opus is automatically upgraded to 1M context with no additional configuration. This applies to both Team Standard and Team Premium seats. On the Anthropic API, Fable 5, Opus 4.8, and Opus 4.7 always run with the 1M window. Sonnet with 1M context is not part of the automatic upgrade and requires [usage credits](https://support.claude.com/en/articles/12429409-extra-usage-for-paid-claude-plans) on every subscription plan, including Max.
| Plan | Opus with 1M context | Sonnet with 1M context |
| - | - | - |
@@ -328,6 +386,7 @@ names** (or equivalent for your API provider), to control the model names that t
| Environment variable | Description |
| - | - |
| `ANTHROPIC_DEFAULT_FABLE_MODEL` | The model to use for `fable`, and the model ID Claude Code recognizes as Fable 5 for [automatic model fallback](#automatic-model-fallback) on third-party providers |
| `ANTHROPIC_DEFAULT_OPUS_MODEL` | The model to use for `opus`, or for `opusplan` when Plan Mode is active. |
| `ANTHROPIC_DEFAULT_SONNET_MODEL` | The model to use for `sonnet`, or for `opusplan` when Plan Mode is not active. |
| `ANTHROPIC_DEFAULT_HAIKU_MODEL` | The model to use for `haiku`, or [background functionality](/en/costs#background-token-usage) |
@@ -380,7 +439,7 @@ These variables take effect on third-party providers such as Bedrock, Vertex AI,
| `ANTHROPIC_DEFAULT_OPUS_MODEL_DESCRIPTION` | Display description for the pinned Opus model in the `/model` picker. Defaults to `Custom Opus model` when not set |
| `ANTHROPIC_DEFAULT_OPUS_MODEL_SUPPORTED_CAPABILITIES` | Comma-separated list of capabilities the pinned Opus model supports |
The same `_NAME`, `_DESCRIPTION`, and `_SUPPORTED_CAPABILITIES` suffixes are available for `ANTHROPIC_DEFAULT_SONNET_MODEL`, `ANTHROPIC_DEFAULT_HAIKU_MODEL`, and `ANTHROPIC_CUSTOM_MODEL_OPTION`.
The same `_NAME`, `_DESCRIPTION`, and `_SUPPORTED_CAPABILITIES` suffixes are available for `ANTHROPIC_DEFAULT_SONNET_MODEL`, `ANTHROPIC_DEFAULT_HAIKU_MODEL`, `ANTHROPIC_DEFAULT_FABLE_MODEL`, and `ANTHROPIC_CUSTOM_MODEL_OPTION`.
Claude Code enables features like [effort levels](#adjust-effort-level) and [extended thinking](#extended-thinking) by matching the model ID against known patterns. Provider-specific IDs such as Bedrock ARNs or custom deployment names often don't match these patterns, leaving supported features disabled. Set `_SUPPORTED_CAPABILITIES` to tell Claude Code which features the model actually supports:
@@ -393,7 +393,7 @@ All metrics and events share these standard attributes:
| `organization.id` | Organization UUID (when authenticated) | Always included when available |
| `user.account_uuid` | Account UUID (when authenticated) | `OTEL_METRICS_INCLUDE_ACCOUNT_UUID` (default: true) |
| `user.account_id` | Account ID in tagged format matching Anthropic admin APIs (when authenticated), such as `user_01BWBeN28...` | `OTEL_METRICS_INCLUDE_ACCOUNT_UUID` (default: true) |
| `user.id` | Anonymous device/installation identifier, generated per Claude Code installation | Always included |
| `user.id` | Random anonymous identifier generated on first run and persisted in `~/.claude.json`. It contains no personal information and is not derived from your Claude account. Deleting the file produces a new unrelated value on next run. | Always included |
| `user.email` | User email address (when authenticated via OAuth) | Always included when available |
| `terminal.type` | Terminal type, such as `iTerm.app`, `vscode`, `cursor`, or `tmux` | Always included when detected |
| Keys from `OTEL_RESOURCE_ATTRIBUTES` | Custom attributes you set, such as `department` or `team.id`. See [Multi-team organization support](#multi-team-organization-support) | `OTEL_METRICS_INCLUDE_RESOURCE_ATTRIBUTES` (default: true) |
@@ -436,7 +436,7 @@ Test each component: run your commands, check agents appear in `/agents`, and ve
| Hooks in `settings.json` | Hooks in `hooks/hooks.json` |
| Must manually copy to share | Install with `/plugin install` |
After migrating, you can remove the original files from `.claude/` to avoid duplicates. The plugin version will take precedence when loaded.
After migrating, remove the original files from `.claude/` to avoid duplicates. Project and user `.claude/agents/` definitions override same-named plugin agents, so the plugin version only takes effect once the originals are removed.
## Next steps
@@ -66,6 +66,8 @@ Each model has its own cache. Switching with [`/model`](/en/model-config#setting
The [`opusplan` model setting](/en/model-config#opusplan-model-setting) resolves to Opus during plan mode and Sonnet during execution, so each plan-mode toggle is a model switch and starts a fresh cache.
[Automatic model fallback](/en/model-config#automatic-model-fallback) on Fable 5 is also a model switch. When a safety classifier flags a request, Claude Code re-runs it on the default Opus model and the session continues there.
### Changing effort level
The cache is keyed by [effort level](/en/model-config#adjust-effort-level) as well as model, so switching with `/effort` means the next request reads the entire conversation history with no cache hits. Once a conversation has started, Claude Code shows a confirmation dialog before applying an effort change that would invalidate the cache. A change that resolves to the same level already in effect, such as setting the model's default explicitly, skips the dialog and keeps the cache.
@@ -17,7 +17,7 @@ Your code's security is paramount. Claude Code is built with security at its cor
Claude Code uses strict read-only permissions by default. When additional actions are needed (editing files, running tests, executing commands), Claude Code requests explicit permission. Users control whether to approve actions once or allow them automatically.
We designed Claude Code to be transparent and secure. For example, we require approval for bash commands before executing them, giving you direct control. This approach enables users and organizations to configure permissions directly.
Claude Code requires approval before running Bash commands that can modify your system. A built-in set of read-only commands such as `ls`, `cat`, and `git status` runs without a prompt. This approach lets users and organizations configure permissions directly.
For detailed permission configuration, see [Permissions](/en/permissions).
@@ -43,7 +43,7 @@ Prompt injection is a technique where an attacker attempts to override or manipu
- **Permission system**: Sensitive operations require explicit approval
- **Context-aware analysis**: Detects potentially harmful instructions by analyzing the full request
- **Input sanitization**: Prevents command injection by processing user inputs
- **Command blocklist**: Blocks risky commands that fetch arbitrary content from the web like `curl` and `wget` by default. When explicitly allowed, be aware of [permission pattern limitations](/en/permissions#tool-specific-permission-rules)
- **Network command approval**: Commands that fetch content from the web such as `curl` and `wget` are not auto-approved by default. They prompt like any other non-read-only Bash command, so you can still approve once or add an explicit allow rule like `Bash(curl *)`. To block them entirely, add them to [`permissions.deny`](/en/permissions#tool-specific-permission-rules)
### Privacy safeguards
@@ -65,7 +65,7 @@ For full details, please review our [Commercial Terms of Service](https://www.an
- **Command injection detection**: Suspicious bash commands require manual approval even if previously allowlisted
- **Fail-closed matching**: Unmatched commands default to requiring manual approval
- **Natural language descriptions**: Complex bash commands include explanations for user understanding
- **Secure credential storage**: API keys and tokens are encrypted. See [Credential Management](/en/authentication#credential-management)
- **Secure credential storage**: API keys and tokens are stored in the macOS Keychain when available, and protected by file permissions on Windows and Linux. See [Credential Management](/en/authentication#credential-management)
**Windows WebDAV security risk**: When running Claude Code on Windows, we recommend against enabling WebDAV or allowing Claude Code to access paths such as `\\*` that may contain WebDAV subdirectories. [WebDAV has been deprecated by Microsoft](https://learn.microsoft.com/en-us/windows/whats-new/deprecated-features#:~:text=The%20Webclient%20\(WebDAV\)%20service%20is%20deprecated) due to security risks. Enabling WebDAV may allow Claude Code to trigger network requests to remote hosts, bypassing the permission system.
@@ -197,7 +197,7 @@ This tolerance applies only to managed settings. User, project, and local settin
| `allowManagedHooksOnly` | (Managed settings only) Only managed hooks, SDK hooks, and hooks from plugins force-enabled in managed settings `enabledPlugins` are loaded. User, project, and all other plugin hooks are blocked. See [Hook configuration](#hook-configuration) | `true` |
| `allowManagedMcpServersOnly` | (Managed settings only) Only `allowedMcpServers` from managed settings are respected. `deniedMcpServers` still merges from all sources. Users can still add MCP servers, but only the admin-defined allowlist applies. See [Managed MCP configuration](/en/managed-mcp) | `true` |
| `allowManagedPermissionRulesOnly` | (Managed settings only) Prevent user and project settings from defining `allow`, `ask`, or `deny` permission rules. Only rules in managed settings apply. See [Managed-only settings](/en/permissions#managed-only-settings) | `true` |
| `alwaysThinkingEnabled` | Enable [extended thinking](/en/model-config#extended-thinking) by default for all sessions. Typically configured via the `/config` command rather than editing directly. To force thinking off regardless of this setting, set [`MAX_THINKING_TOKENS=0`](/en/env-vars) in `env`. On [third-party providers](/en/third-party-integrations) this omits the `thinking` parameter instead, and adaptive-reasoning models may still think | `true` |
| `alwaysThinkingEnabled` | Enable [extended thinking](/en/model-config#extended-thinking) by default for all sessions. Typically configured via the `/config` command rather than editing directly. To force thinking off regardless of this setting, set [`MAX_THINKING_TOKENS=0`](/en/env-vars) in `env`, which disables thinking on the Anthropic API except on Fable 5, which cannot have thinking turned off. On [third-party providers](/en/third-party-integrations) this omits the `thinking` parameter instead, and adaptive-reasoning models may still think | `true` |
| `apiKeyHelper` | Custom script, to be executed in `/bin/sh`, to generate an auth value. This value will be sent as `X-Api-Key` and `Authorization: Bearer` headers for model requests. Set the refresh interval with [`CLAUDE_CODE_API_KEY_HELPER_TTL_MS`](/en/env-vars) | `/bin/generate_temp_api_key.sh` |
| `attribution` | Customize attribution for git commits and pull requests. See [Attribution settings](#attribution-settings) | `{"commit": "🤖 Generated with Claude Code", "pr": ""}` |
| `autoMemoryDirectory` | Custom directory for [auto memory](/en/memory#storage-location) storage. Accepts an absolute path or a `~/`-prefixed path. From project or local settings, this is honored only after you accept the workspace trust dialog, since a cloned repository can supply this file | `"~/my-memory-dir"` |
@@ -71,6 +71,8 @@ Claude Code includes additional helper agents for specific tasks. These are typi
| statusline-setup | Sonnet | When you run `/statusline` to configure your status line |
| claude-code-guide | Haiku | When you ask questions about Claude Code features |
Built-in subagents are always registered in interactive sessions. To block a specific built-in type, add it to `permissions.deny` as shown in [Disable specific subagents](#disable-specific-subagents). To prevent Claude from delegating to any subagent, deny the `Agent` tool itself with [`permissions.deny`](/en/permissions#tool-specific-permission-rules). In [non-interactive mode](/en/headless) and the [Agent SDK](/en/agent-sdk/overview), set [`CLAUDE_AGENT_SDK_DISABLE_BUILTIN_AGENTS=1`](/en/env-vars) to remove all built-in types and supply only your own.
Beyond these built-in subagents, you can create your own with custom prompts, tool restrictions, permission modes, hooks, and skills. The following sections show how to get started and customize subagents.
## Quickstart: create your first subagent
@@ -229,7 +231,7 @@ The following fields can be used in the YAML frontmatter. Only `name` and `descr
| `description` | Yes | When Claude should delegate to this subagent |
| `tools` | No | [Tools](#available-tools) the subagent can use. Inherits all tools if omitted. To preload Skills into context, use the `skills` field rather than listing `Skill` here |
| `disallowedTools` | No | Tools to deny, removed from inherited or specified list |
| `model` | No | [Model](#choose-a-model) to use: `sonnet`, `opus`, `haiku`, a full model ID (for example, `claude-opus-4-8`), or `inherit`. Defaults to `inherit` |
| `model` | No | [Model](#choose-a-model) to use: `sonnet`, `opus`, `haiku`, `fable`, a full model ID (for example, `claude-opus-4-8`), or `inherit`. Defaults to `inherit` |
| `permissionMode` | No | [Permission mode](#permission-modes): `default`, `acceptEdits`, `auto`, `dontAsk`, `bypassPermissions`, or `plan`. Ignored for [plugin subagents](#choose-the-subagent-scope) |
| `maxTurns` | No | Maximum number of agentic turns before the subagent stops |
| `skills` | No | [Skills](/en/skills) to preload into the subagent's context at startup. The full skill content is injected, not just the description. Subagents can still invoke unlisted project, user, and plugin skills through the Skill tool |
@@ -246,7 +248,7 @@ The following fields can be used in the YAML frontmatter. Only `name` and `descr
The `model` field controls which [AI model](/en/model-config) the subagent uses:
- **Model alias**: Use one of the available aliases: `sonnet`, `opus`, or `haiku`
- **Model alias**: Use one of the available aliases: `sonnet`, `opus`, `haiku`, or `fable`
- **Full model ID**: Use a full model ID such as `claude-opus-4-8` or `claude-sonnet-4-6`. Accepts the same values as the `--model` flag
- **inherit**: Use the same model as the main conversation
- **Omitted**: If not specified, defaults to `inherit` (uses the same model as the main conversation)
@@ -535,6 +535,10 @@ Git for Windows is optional. Claude Code uses the [PowerShell tool](/en/tools-re
If your Git is installed somewhere else, find the path by running `where.exe git` in PowerShell and use the `bin\bash.exe` path from that directory.
**If the path is correct and the file exists** but Claude Code still reports it as not found, endpoint security software such as AppLocker, Group Policy software restriction policies, or EDR agents may be interfering. On versions before v2.1.116, Claude Code spawned a child process (`cmd.exe`) to verify the path, which these policies can block — a common signal is that `cmd.exe /c dir "C:\Program Files\Git\bin\bash.exe"` works when you run it directly in PowerShell but fails silently when launched by `claude.exe`.
Claude Code v2.1.116 and later check the filesystem directly, so update first. If the error persists on a current version, ask your IT team to allowlist `claude.exe` and the processes it spawns, including `cmd.exe` and `bash.exe`, in your endpoint protection policy.
### Claude Code does not support 32-bit Windows
Windows includes two PowerShell entries in the Start menu: `Windows PowerShell` and `Windows PowerShell (x86)`. The x86 entry runs as a 32-bit process and triggers this error even on a 64-bit machine. To check which case you're in, run this in the same window that produced the error:
@@ -28,7 +28,7 @@ ZDR is enabled on a per-organization basis. Each new organization requires ZDR t
### What ZDR covers
ZDR covers model inference calls made through Claude Code on Claude for Enterprise. When you use Claude Code in your terminal, the prompts you send and the responses Claude generates are not retained by Anthropic. This applies regardless of which Claude model is used.
ZDR covers model inference calls made through Claude Code on Claude for Enterprise. When you use Claude Code in your terminal, the prompts you send and the responses Claude generates are not retained by Anthropic. This applies to every model available to ZDR organizations. Some models require data retention and are not available under ZDR; see [Model availability under ZDR](#model-availability-under-zdr).
### What ZDR does not cover
@@ -56,6 +56,12 @@ These features are blocked in the backend regardless of client-side display. If
Future features may also be disabled if they require storing prompts or completions.
### Model availability under ZDR
Claude Fable 5 is not available for organizations with zero data retention enabled. This model class [requires data retention](https://platform.claude.com/docs/en/manage-claude/api-and-data-retention#model-specific-data-retention-requirements), so requests from ZDR organizations cannot be served by it. The model is either absent from the `/model` picker for ZDR organizations or shown as disabled with a notice that disabling ZDR is required, and the server rejects requests for it regardless of client configuration.
Other models remain available under ZDR. Fable 5 is not the default model, and the `best` alias, which resolves to Fable 5 where it is available, resolves to Opus for organizations where it is not, including ZDR organizations.
## Data retention for policy violations
Even with ZDR enabled, Anthropic may retain data where required by law or to address Usage Policy violations. If a session is flagged for a policy violation, Anthropic may retain the associated inputs and outputs for up to 2 years, consistent with Anthropic's standard ZDR policy.