System Configuration

10s 300s
500 5000
Show Exhausted Models Display models even if they have 0% remaining quota.
Compact Mode Reduce padding in tables for higher information density.
Show Config Warning Display a warning on Dashboard when Claude CLI config doesn't match preset.
Settings below directly modify ~/.claude/settings.json. Restart Claude CLI to apply.
Proxy Paid

Claude CLI is using the official Anthropic API

All proxy configuration has been removed. Claude CLI uses your Anthropic subscription directly.

Switch to Proxy mode to configure model routing and presets.

Select a preset to load it. Click "Apply to Claude CLI" to save changes.

ANTHROPIC_BASE_URL
ANTHROPIC_AUTH_TOKEN
▼
▼
DEFAULT MODEL ALIASES
▼
▼
▼
Experimental MCP CLI Enables experimental MCP integration for reliable tool usage with reduced context consumption.
Gemini 1M Context Mode Appends [1m] suffix to Gemini models for 1M context window support. ⚠ Large context may reduce Gemini performance.

Confirm Restore

Are you sure you want to restore Claude CLI to default settings? This will remove proxy configuration.

Unsaved Changes

Your current configuration doesn't match any saved preset.

Save Preset

Save the current configuration as a reusable preset.

Configure model visibility, pinning, and request mapping.
Model mapping: server-side redirection. Claude Code users: see 'Claude CLI' tab for client-side setup.
Show Hidden Models
Model ID Mapping (Target Model ID) Actions
NO MODELS DETECTED

WebUI Password

Authentication for accessing this dashboard

Select a preset and click the download icon to apply it. Built-in presets (🔒) cannot be deleted.

Save Server Preset
Common Settings
Developer Mode
Enables debug logging, log export, and strategy health inspector
Screenshot Mode Hide email addresses for safe screen sharing.
Debug Logging Show verbose debug messages in the log viewer.
Log Export Show export button in the log viewer toolbar.
Health Inspector Show strategy health inspector in the accounts view.
Placeholder Data Inject dummy accounts and quotas for UI testing.
Include Real Accounts Show real accounts alongside placeholder data.
Persist Token Cache Save OAuth tokens to disk for faster restarts
Maximum number of Google accounts allowed
Account Selection Strategy
Selection Strategy How accounts are selected for requests
Advanced Settings Settings managed via config.json
Network Retry Settings
Rate Limiting & Timeouts

Fallback cooldown when API doesn't provide a reset time.

If all accounts are rate-limited longer than this, error immediately.

Quota Protection

Switch accounts when quota drops below this level. Per-account overrides take priority.

Request Throttling Experimental
Enable Request Throttling Add delay before each Google API request to avoid rate limits.

Milliseconds to wait before each API call (token refresh, quota check, messages, etc).

Error Handling Tuning

Prevents concurrent retry storms.

Failures before extended cooldown.

Applied after max consecutive failures.

Retries before switching accounts.

All changes are saved automatically. Some settings may require server restart.

Change WebUI Password