Skip to content

luw2007/llm-balance

BranchesTags

Repository files navigation

LLM Balance Checker

A unified CLI tool for monitoring costs and usage across multiple LLM platforms

Key Features

  • 🔑 Multiple Authentication: API Key, browser cookie, official SDK support
  • 🌐 27 Platforms Supported: Official platforms (DeepSeek, Moonshot, Volcengine, Aliyun, Tencent, Zhipu, SiliconFlow, OpenAI, Anthropic, Google) + third-party relays (FoxCode, DuckCoding, PackyCode, 88Code, 88996, AICoding, YouAPI, CSMindAI, YesVg, Cubence, DawClaudeCode, Magic666, Jimiai) + relay platforms (OneAPI, APIProxy, FastGPT, MiniMax)
  • 💰 Real-time Balance & Spent: Track both current balance and actual spending
  • 📊 Flexible Output: Table, JSON, Markdown, and total-only formats
  • 💱 Multi-Currency: Automatic conversion between CNY, USD, EUR, and more
  • 🎯 Token Monitoring: Detailed token usage for supported platforms
  • 🛡️ Fault Tolerant: Single platform failures won't break the entire tool
  • ⚙️ Easy Configuration: Simple setup with environment variables
  • 🔒 Independent Configuration: Special platforms use separate config files to avoid global pollution
  • ⚡ High Performance: Multi-threaded concurrent checking (5 workers) for fast multi-platform queries

Quick Start

Installation

# Clone the repository
git clone <repository-url>
cd llm_balance

# Install dependencies (includes Volcengine SDK)
pip install -r requirements.txt

# Install CLI tool
pip install -e .

# Verify installation
llm-balance --help
llm-balance rates

Platform Setup

Choose the platforms you want to use and configure their authentication:

API Key Platforms (Quick Setup)

# DeepSeek (API Key)
export DEEPSEEK_API_KEY="sk-your-deepseek-api-key"

# Moonshot (API Key)
export MOONSHOT_API_KEY="sk-your-moonshot-api-key"

# SiliconFlow (API Key)
export SILICONFLOW_API_KEY="your-siliconflow-api-key"

SDK Platforms (Enterprise Setup)

# Volcengine (Official SDK)
export VOLCENGINE_ACCESS_KEY="your-volcengine-access-key"
export VOLCENGINE_SECRET_KEY="your-volcengine-secret-key"

# Aliyun (Official SDK)
export ALIYUN_ACCESS_KEY_ID="your-aliyun-access-key-id"
export ALIYUN_ACCESS_KEY_SECRET="your-aliyun-access-key-secret"

# Tencent Cloud (SDK)
export TENCENT_API_KEY="your-tencent-secret-id:your-tencent-secret-key"

Browser-based Platforms

# Zhipu AI (Requires browser login)
# 1. Login to https://open.bigmodel.cn
# 2. Set global browser for cookie extraction
llm-balance set-browser chrome
# Or override per command: llm-balance cost --browser=chrome

Special Platforms with Independent Configuration

# DuckCoding (Requires independent configuration)
# Method 1: Environment variable
export DUCKCODING_API_USER_ID="your_user_id"

# Method 2: Separate config file
llm-balance platform_config duckcoding api_user_id your_user_id

# Method 3: Manual config file creation
cat > ~/.llm_balance/duckcoding_config.yaml << EOF
api_user_id: your_user_id
EOF

# Then login to https://duckcoding.com and run:
llm-balance cost --platform=duckcoding

# PackyCode (Requires independent configuration)
# Method 1: Environment variable
export PACKYCODE_API_USER_ID="your_user_id"

# Method 2: Separate config file
llm-balance platform_config packycode api_user_id your_user_id

# Method 3: Manual config file creation
cat > ~/.llm_balance/packycode_config.yaml << EOF
api_user_id: your_user_id
EOF

# Then login to https://packyapi.com and run:
llm-balance cost --platform=packycode

# CSMindAI (Requires independent configuration)
# Method 1: Environment variable
export CSMINDAI_API_USER_ID="your_user_id"

# Method 2: Separate config file
llm-balance platform_config csmindai api_user_id your_user_id

# Method 3: Manual config file creation
cat > ~/.llm_balance/csmindai_config.yaml << EOF
api_user_id: your_user_id
EOF

# Then login to https://api.csmindai.com and run:
llm-balance cost --platform=csmindai

# YouAPI (Requires independent configuration)
# Method 1: Environment variable
export YOUAPI_NEW_API_USER="your_user_id"

# Method 2: Separate config file
llm-balance platform_config yourapi api_user_id your_user_id

# Method 3: Manual config file creation
cat > ~/.llm_balance/youapi_config.yaml << EOF
api_user_id: your_user_id
EOF

# Then login to https://yourapi.cn and run:
llm-balance cost --platform=yourapi

# Cubence (Requires independent configuration)
# Method 1: Environment variable
export CUBENCE_TOKEN="your_token"

# Method 2: Separate config file
llm-balance platform_config cubence token "your_token"

# Method 3: Manual config file creation
cat > ~/.llm_balance/cubence_config.yaml << EOF
token: your_token
EOF

# Then run:
llm-balance cost --platform=cubence

# 88996.cloud (Requires independent configuration)
# Method 1: Environment variable
export CLOUD88996_API_USER_ID="your_user_id"

# Method 2: Manual config file creation
cat > ~/.llm_balance/88996_config.yaml << EOF
api_user_id: your_user_id
EOF

# Then login to https://88996.cloud and run:
llm-balance cost --platform=88996

First Use

# Test a single platform (choose one you configured)
llm-balance cost --platform=deepseek

# Check all configured platforms
llm-balance cost

# View results in different formats
llm-balance cost --format=table    # Clean console output
llm-balance cost --format=json     # Machine-readable
llm-balance cost --format=total    # Just the totals

Quick Troubleshooting

If you encounter issues, try these steps:

# Check if environment variables are set
echo $DEEPSEEK_API_KEY
echo $VOLCENGINE_ACCESS_KEY

# Test individual platform with debug output
llm-balance cost --platform=deepseek --format=json

# Verify configuration
llm-balance list

# Check browser setting for cookie-based platforms
llm-balance set-browser chrome

# Test network connectivity
curl -I https://api.deepseek.com

Common Issues:

  • API Key not found: Make sure environment variables are set correctly
  • Browser authentication failed: Ensure you're logged into the platform website
  • Permission denied: Check API key permissions and account status
  • Network timeout: Verify internet connection and platform API status

Usage

Essential Commands

Balance & Spent Monitoring

# Check all configured platforms
llm-balance cost

# Check specific platform
llm-balance cost --platform=deepseek

# Multiple platforms (comma-separated)
llm-balance cost --platform=volcengine,aliyun

# Browser override for cookie-based platforms
llm-balance cost --platform=zhipu --browser=chrome

# Output formats
llm-balance cost --format=table     # Clean console view (default)
llm-balance cost --format=json      # Machine-readable
llm-balance cost --format=total     # Just the totals

# Currency conversion
llm-balance cost --currency=USD     # Convert to USD
llm-balance cost --currency=EUR     # Convert to EUR

💡 New Feature: The cost command now displays both current balance and spent amount for all platforms, providing a complete financial overview of your LLM usage.

Token Usage Monitoring

# Check token usage for supported platforms
llm-balance package

# Specific platform and model
llm-balance package --platform=volcengine --model=deepseek-r1

# Multiple platforms
llm-balance package --platform=volcengine,zhipu

# Different output formats
llm-balance package --format=table   # Console view
llm-balance package --format=json    # Machine-readable
llm-balance package --show-all       # Include inactive entries

# Show subscription lifecycle information (for 88code, FoxCode, Moonshot)
llm-balance package --show-expiry      # Show expiry dates
llm-balance package --show-reset       # Show reset counts
llm-balance package --show-reset-time  # Show reset times
llm-balance package --show-expiry --show-reset --show-reset-time  # Show all

Note: Token monitoring is available for Volcengine, Zhipu, DuckCoding, CSMindAI, YouAPI, 88Code, 88996, DawClaudeCode, Magic666, Jimiai, FoxCode, and Moonshot platforms

Subscription Features: For platforms with subscription-based billing (88Code, FoxCode), the tool automatically distinguishes between:

  • Pay-per-use packages: Calculated by actual usage
  • Subscription packages: Calculated by time depreciation with reset information

By default the package command hides entries marked as inactive. Pass --show-all to include them across all output formats.

Coding Plan Usage

Check coding plan limits for Claude Code compatible platforms (Volcengine, Zhipu, ChatGPT):

# Check all platforms
llm-balance plan

# Check specific platform
llm-balance plan --platform=volcengine
llm-balance plan --platform=chatgpt

# Use specific browser for cookies
llm-balance plan --browser=doubao

Note: Coding Plan monitoring requires platform-specific authentication:

  • Volcengine: Browser cookies (login to console.volcengine.com)
  • Zhipu AI: Set ZHIPU_AUTH_TOKEN or ANTHROPIC_AUTH_TOKEN environment variable
  • ChatGPT: Read local ~/.codex/sessions/**/rollout-*.jsonl logs generated by Codex CLI sessions (codex alias supported)
  • ChatGPT Review Quota: If review has dedicated limits in logs, show review_session/review_weekly; otherwise fallback to ChatGPT main quota when review activity exists

Quick Reference

Command Purpose Example
llm-balance cost Check balance & spent llm-balance cost --platform=volcengine
llm-balance package Check token usage llm-balance package --platform=zhipu
llm-balance plan Check coding plan limits llm-balance plan --platform=volcengine
llm-balance list List platforms llm-balance list
llm-balance enable Enable platform llm-balance enable deepseek
llm-balance disable Disable platform llm-balance disable tencent

Platform Management

# List all platforms and their status
llm-balance list

# Enable platform
llm-balance enable moonshot

# Disable platform
llm-balance disable moonshot

# Switch authentication method (SDK vs Cookie)
llm-balance config volcengine auth_type sdk    # Use official SDK
llm-balance config volcengine auth_type cookie  # Use browser cookies

# Independent configuration for special platforms
llm-balance platform_config duckcoding         # View DuckCoding config
llm-balance platform_config duckcoding api_user_id your_user_id  # Set user ID

Configuration Management

# View complete platform configuration
llm-balance config deepseek

# View specific configuration item
llm-balance config deepseek api_url

# Set configuration item (supports automatic type conversion)
llm-balance config deepseek enabled true
llm-balance config deepseek timeout 30

Advanced Usage Examples

Balance Checking

# View total balance in USD for all platforms
llm-balance cost --currency=USD --format=total

# Get detailed EUR balance for specific platform
llm-balance cost --platform=deepseek --currency=EUR --format=table

# Check multiple platforms at once
llm-balance cost --platform=volcengine,aliyun,deepseek --format=table
llm-balance cost --platform="openai, deepseek, moonshot" --currency=USD --format=total

# Use custom exchange rates and output as JSON
LLM_BALANCE_RATES='{"USD": 7.5}' llm-balance cost --currency=USD --format=json

# Combine multiple parameters
llm-balance cost --platform=openai --browser=chrome --currency=GBP --format=markdown

# Test Volcengine SDK authentication
export VOLCENGINE_ACCESS_KEY="your_access_key"
export VOLCENGINE_SECRET_KEY="your_secret_key"
llm-balance cost --platform=volcengine --format=table

# Test Chinese platforms
export TENCENT_API_KEY="your_tencent_api_key"
llm-balance cost --platform=tencent --format=table

# Test cookie authentication platforms
# Zhipu requires login to open.bigmodel.cn
llm-balance cost --platform=zhipu --browser=chrome

Token Usage Monitoring

# Check token usage across all supported platforms
llm-balance package

# Check token usage for specific model on Volcengine
llm-balance package --platform=volcengine --model=deepseek-r1

# Check token usage for Zhipu AI with specific model
llm-balance package --platform=zhipu --model=glm-4-plus

# Compare token usage across multiple platforms
llm-balance package --platform=volcengine,zhipu --format=table

# Get detailed JSON output for token usage
llm-balance package --platform=volcengine --format=json

# Filter models by partial name matching
llm-balance package --platform=volcengine --model=deepseek
llm-balance package --platform=zhipu --model=glm-4

# Check token usage in different currencies
llm-balance package --currency=USD --format=table

Exchange Rate Features

Overview

Supports automatic exchange rate conversion to convert balances in different currencies to a target currency for total calculation. Supports multiple currency types including CNY, USD, EUR, etc.

Default Exchange Rates

System includes following default rates (relative to CNY):

  • CNY: 1.0
  • USD: 7.2
  • EUR: 7.8
  • GBP: 9.1
  • JPY: 0.048
  • KRW: 0.0054
  • Points: 0.01

Viewing and Managing Exchange Rates

# View current exchange rates
llm-balance rates

# Custom exchange rates (session level)
LLM_BALANCE_RATES='{"USD": 7.5}' llm-balance cost

# Set multiple exchange rates
LLM_BALANCE_RATES='{"USD": 7.5, "EUR": 8.0}' llm-balance cost

Currency Conversion Rules

  • Flexible Conversion: Supports conversion between any currencies
  • Total Display: Table and total formats show totals in specified currency
  • Original Data: JSON and Markdown formats preserve original currency display
  • Default Currency: Uses CNY as default output currency
  • Custom Rates: Set via LLM_BALANCE_RATES environment variable

Supported Currencies

  • CNY (Chinese Yuan)
  • USD (US Dollar)
  • EUR (Euro)
  • GBP (British Pound)
  • JPY (Japanese Yen)
  • KRW (Korean Won)
  • Points (Points)

Configuration File

Configuration file is located at ~/.llm_balance/platforms.yaml, automatically created on first run:

# Global browser configuration for all cookie-based platforms
browser: chrome

platforms:
  deepseek:
    api_url: "https://api.deepseek.com/user/balance"
    method: "GET"
    auth_type: "bearer_token"
    env_var: "DEEPSEEK_API_KEY"
    balance_path: ["balance_infos", "0", "total_balance"]
    currency_path: ["balance_infos", "0", "currency"]
    enabled: true
    headers:
      User-Agent: "Mozilla/5.0 (Macintosh; Intel Mac OS X 10_15_7) AppleWebKit/537.36"
      Accept: "application/json"
  
  openai:
    api_url: "https://api.openai.com/v1/organization/costs"
    method: "GET"
    auth_type: "bearer_token"
    env_var: "OPENAI_ADMIN_KEY"
    balance_path: ["data", "0", "results", "0", "amount", "value"]
    currency_path: ["data", "0", "results", "0", "amount", "currency"]
    enabled: true
    params:
      start_time: 1730419200
      limit: 1

Supported Platforms

🌍 International Platforms (3)

Platform Authentication Status Description Token Usage Spent Tracking
OpenAI Admin API ✅ Requires OPENAI_ADMIN_KEY ❌ Not Available ✅ Full Support
Anthropic Admin API ✅ Requires ANTHROPIC_ADMIN_KEY ❌ Not Available ✅ Full Support
Google API Key ✅ Requires GOOGLE_API_KEY ❌ Not Available ✅ Full Support

🇨🇳 Chinese Platforms (7)

Platform Authentication Status Description Token Usage Spent Tracking
DeepSeek API Key ✅ Requires DEEPSEEK_API_KEY ❌ Not Available ❌ Not Supported
Moonshot API Key ✅ Requires MOONSHOT_API_KEY ❌ Not Available ❌ Not Supported
Volcengine SDK/Cookie ✅ Requires VOLCENGINE_ACCESS_KEY + SECRET_KEY or browser login ✅ Full Support ✅ Full Support
Aliyun Official SDK ✅ Requires ALIYUN_ACCESS_KEY_ID + SECRET_KEY ❌ Not Available ✅ Full Support
Tencent SDK ✅ Requires TENCENT_API_KEY (SecretId:SecretKey) ❌ Not Available ✅ Available
Zhipu Cookie ✅ Requires login to https://open.bigmodel.cn ✅ Full Support ✅ Full Support
SiliconFlow API Key ✅ Requires SILICONFLOW_API_KEY ❌ Not Available ✅ Full Support

🔄 Third-Party Relay Platforms (9)

Platform Authentication Status Description Token Usage Spent Tracking Independent Config
FoxCode Cookie ✅ Relay service with dashboard access ✅ Full Support ✅ Full Support ❌ No
DuckCoding Cookie ✅ Relay service with token packages ✅ Full Support ✅ Full Support ✅ Yes
88Code Console Token ✅ Relay service with subscription packages ✅ Full Support ✅ Full Support ✅ Yes
88996.cloud Cookie ✅ Relay service with quota system ✅ Full Support ✅ Full Support ✅ Yes
DawClaudeCode Cookie ✅ Relay service with quota system ✅ Full Support ✅ Full Support ✅ Yes
Magic666 Cookie ✅ Relay service with quota system ✅ Full Support ✅ Full Support ✅ Yes
Jimiai Cookie ✅ Relay service with quota system ✅ Full Support ✅ Full Support ✅ Yes
AICoding Cookie ✅ Relay service with credits system ✅ Full Support ❌ Not Available ❌ No
YourAPI Cookie ✅ Relay service with quota system ✅ Full Support ✅ Full Support ✅ Yes

📊 Platform Status Summary

Production-Ready (17 platforms): All platforms listed above are fully tested and ready for production use.

Independent Configuration: DuckCoding, 88Code, 88996, DawClaudeCode, Magic666, Jimiai, and YourAPI use separate configuration files to avoid polluting global settings.

Development Status: Additional platforms (Azure OpenAI, Lingyi, MiniMax) are available in the dev branch and under active development.

Token Usage Support

Some platforms provide token usage monitoring capabilities:

Platform Token Usage Support Models Available
Volcengine ✅ Full Support deepseek-r1, deepseek-v3, doubao-pro-32k, doubao-pro-128k, etc.
Zhipu AI ✅ Full Support glm-4-plus, glm-4, glm-3-turbo, cogview-3, etc.
DeepSeek ❌ Not Available -
Moonshot ❌ Not Available -
Aliyun ❌ Not Available -
Tencent ❌ Not Available -
SiliconFlow ❌ Not Available -

Spent Amount Tracking

Real-time spent amount tracking across supported platforms:

Platform Spent Tracking Data Source Accuracy
Volcengine ✅ Full Support Official API High
Aliyun ✅ Full Support BSS Transaction API High
Zhipu AI ✅ Full Support Billing API High
Tencent Cloud ✅ Available Billing API Medium
SiliconFlow ✅ Full Support Billing API High
DeepSeek ❌ Not Supported N/A N/A
Moonshot ❌ Not Supported N/A N/A

Note: Platforms marked as "Not Supported" for spent tracking will display "-" instead of a numeric value to clearly distinguish between unsupported functionality and zero spending.

Authentication Methods

🔑 API Key Authentication

For platforms providing API interfaces:

# DeepSeek
export DEEPSEEK_API_KEY="sk-xxxxxxxxxxxxxxxxxxxxxxxx"

# OpenAI (currently not supported for balance queries)
# export OPENAI_ADMIN_KEY="sk-admin-xxxxxxxxxxxxxxxxxxxxxxxx"

# Aliyun (requires both access key ID and secret)
export ALIYUN_ACCESS_KEY_ID="your_access_key_id"
export ALIYUN_ACCESS_KEY_SECRET="your_access_key_secret"

# Moonshot
export MOONSHOT_API_KEY="sk-xxxxxxxxxxxxxxxxxxxxxxxx"

# SiliconFlow
export SILICONFLOW_API_KEY="your_siliconflow_api_key"

🍪 Cookie Authentication

For platforms requiring browser sessions:

  1. Login to the corresponding platform website
  2. Tool automatically retrieves authentication cookies from browser
  3. Supports Chrome, Firefox, Arc, Brave and other mainstream browsers

🌐 Special Services

Some platforms use special access methods:

Third-Party Relay: FoxCode

FoxCode is a cookie-authenticated relay with dashboard-based package and cost information.

  • Auth: Browser cookie on foxcode.rjj.cc. Read auth_token and query https://foxcode.rjj.cc/api/user/dashboard with Authorization: Bearer <token>.
  • package: Uses data.subscription.active entries.
    • Total = plan quotaLimit
    • Remaining = quotaRemaining (fallback to plan duration)
    • Used = Total - Remaining
    • Package column shows plan.name
  • cost: Balance is shown as - (no top-up on this relay); Spent = sum of data.subscription.history[*].plan.price in CNY.

Examples:

llm-balance package --platform=foxcode
llm-balance cost --platform=foxcode
# Specify browser if needed
llm-balance package --platform=foxcode --browser=chrome
llm-balance cost --platform=foxcode --browser=chrome

Third-Party Relay: DuckCoding

DuckCoding is a cookie-authenticated relay with token-based package and cost information, using independent configuration.

  • Auth: Browser cookie on duckcoding.com. Query https://duckcoding.com/api/user/self with new-api-user header.
  • Configuration: Requires api_user_id setting via environment variable or separate config file.
  • package: Uses data.quota and data.used_quota from user info.
    • Total = quota (in tokens)
    • Used = used_quota (in tokens)
    • Remaining = Total - Used
    • Package column shows "DuckCoding 按量计费(不到期)"
  • cost: Balance and spent calculated from quota data.
    • Balance = quota / 500000 (in CNY)
    • Spent = used_quota / 500000 (in CNY)

Configuration Options:

# Method 1: Environment variable
export DUCKCODING_API_USER_ID="your_user_id"

# Method 2: CLI command
llm-balance platform_config duckcoding api_user_id your_user_id

# Method 3: Manual config file
cat > ~/.llm_balance/duckcoding_config.yaml << EOF
api_user_id: your_user_id
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=duckcoding
# Check token usage
llm-balance package --platform=duckcoding
# View configuration
llm-balance platform_config duckcoding
# Configure user ID
llm-balance platform_config duckcoding api_user_id your_user_id

Third-Party Relay: 88Code

88Code is a console token-authenticated relay with subscription-based packages and cost information.

  • Auth: Console token via environment variable or separate config file.
  • Configuration: Requires console_token setting.
  • package: Uses subscription data from https://www.88code.org/admin-api/cc-admin/system/subscription/my.
    • Total = sum of all subscription costs (active + inactive)
    • Balance = calculated based on usage ratio of active subscriptions
    • Spent = Total cost - Balance
    • Package column shows subscription features with status display
  • Status: Shows "active" or "inactive" based on subscription isActive flag

Configuration Options:

# Method 1: Environment variable
export CODE88_CONSOLE_TOKEN="your_console_token"

# Method 2: Manual config file
cat > ~/.llm_balance/88code_config.yaml << EOF
console_token: your_console_token
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=88code
# Check token usage with status display
llm-balance package --platform=88code

Third-Party Relay: YourAPI

YourAPI is a cookie-authenticated relay with simple quota-based balance and spent calculation.

  • Auth: Browser cookie on yourapi.cn with new-api-user header.
  • Configuration: Requires api_user_id setting via environment variable or separate config file.
  • package: Uses quota data from https://yourapi.cn/api/user/self.
    • Total = quota (in tokens, converted to CNY by dividing by 500000)
    • Used = used_quota (in tokens, converted to CNY by dividing by 500000)
    • Remaining = Total - Used
    • Package column shows user group information
    • Status: Shows "active" or "inactive" based on user status
  • cost: Balance and spent calculated from quota data with currency conversion.

Configuration Options:

# Method 1: Environment variable
export YOURAPI_NEW_API_USER="your_user_id"

# Method 2: Manual config file
cat > ~/.llm_balance/yourapi_config.yaml << EOF
api_user_id: "your_user_id"
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=yourapi
# Check token usage
llm-balance package --platform=yourapi
# View configuration
llm-balance platform_config yourapi

Third-Party Relay: 88996.cloud

88996.cloud is a cookie-authenticated relay with quota-based balance and package information.

  • Auth: Browser cookie on 88996.cloud with rix-api-user header.
  • Configuration: Requires api_user_id setting via environment variable or separate config file.
  • package: Uses quota data from https://88996.cloud/api/user/self.
    • Total = quota + bonus_quota (in tokens)
    • Used = used_quota (in tokens)
    • Remaining = Total - Used
  • cost: Balance and spent calculated from quota data.
    • Balance = (quota + bonus_quota) / 500000 (in CNY)
    • Spent = used_quota / 500000 (in CNY)

Configuration Options:

# Method 1: Environment variable
export CLOUD88996_API_USER_ID="your_user_id"

# Method 2: Manual config file
cat > ~/.llm_balance/88996_config.yaml << EOF
api_user_id: "your_user_id"
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=88996
# Check token usage
llm-balance package --platform=88996

Third-Party Relay: DawClaudeCode

DawClaudeCode is a cookie-authenticated relay with quota-based balance and package information.

  • Auth: Browser cookie on dawclaudecode.com with new-api-user header.
  • Configuration: Requires api_user_id setting via environment variable or separate config file.
  • package: Uses quota data from https://dawclaudecode.com/api/user/self.
    • Total = quota + bonus_quota (in tokens)
    • Used = used_quota (in tokens)
    • Remaining = Total - Used
  • cost: Balance and spent calculated from quota data.
    • Balance = (quota + bonus_quota) / 500000 (in CNY)
    • Spent = used_quota / 500000 (in CNY)

Configuration Options:

# Method 1: Environment variable
export DAWCLAUDECODE_API_USER_ID="your_user_id"

# Method 2: Manual config file
cat > ~/.llm_balance/dawclaudecode_config.yaml << EOF
api_user_id: "your_user_id"
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=dawclaudecode
# Check token usage
llm-balance package --platform=dawclaudecode

Third-Party Relay: Magic666

Magic666 is a cookie-authenticated relay with quota-based balance and package information.

  • Auth: Browser cookie on magic666.top with new-api-user header.
  • Configuration: Requires api_user_id setting via environment variable or separate config file.
  • package: Uses quota data from https://magic666.top/api/user/self.
    • Total = quota + bonus_quota (in tokens)
    • Used = used_quota (in tokens)
    • Remaining = Total - Used
  • cost: Balance and spent calculated from quota data.
    • Balance = (quota + bonus_quota) / 500000 (in CNY)
    • Spent = used_quota / 500000 (in CNY)

Configuration Options:

# Method 1: Environment variable
export MAGIC666_API_USER_ID="your_user_id"

# Method 2: Manual config file
cat > ~/.llm_balance/magic666_config.yaml << EOF
api_user_id: "your_user_id"
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=magic666
# Check token usage
llm-balance package --platform=magic666

Third-Party Relay: Jimiai

Jimiai is a cookie-authenticated relay with quota-based balance and package information.

  • Auth: Browser cookie on jimiai.ai with new-api-user header.
  • Configuration: Requires api_user_id setting via environment variable or separate config file.
  • package: Uses quota data from https://jimiai.ai/api/user/self.
    • Total = quota (in tokens)
    • Used = used_quota (in tokens)
    • Remaining = Total - Used
  • cost: Balance and spent calculated from quota data.
    • Balance = quota / 500000 (in CNY)
    • Spent = used_quota / 500000 (in CNY)

Configuration Options:

# Method 1: Environment variable
export JIIMIAI_API_USER_ID="your_user_id"

# Method 2: Manual config file
cat > ~/.llm_balance/jimiai_config.yaml << EOF
api_user_id: "your_user_id"
EOF

Examples:

# Check balance and spent
llm-balance cost --platform=jimiai
# Check token usage
llm-balance package --platform=jimiai

Third-Party Relay: AICoding

AICoding is a cookie-authenticated relay with credits-based balance information.

  • Auth: Browser cookie on aicoding.sh.
  • package: Uses credits data from https://aicoding.sh/api/user-credits/permanent.
    • Total = personal_credits (in credits)
    • Remaining = personal_credits (in credits)
  • cost: Balance calculated from credits data.
    • Balance = personal_credits / 100 (in CNY)
    • Spent = Not available (returns "-")

Examples:

# Login to https://aicoding.sh first, then:
# Check balance
llm-balance cost --platform=aicoding
# Check credits usage
llm-balance package --platform=aicoding

Browser Support

🌍 Global Browser Configuration

All cookie-based platforms use a single global browser configuration:

# Set global browser (affects all cookie platforms)
llm-balance set-browser chrome
llm-balance set-browser firefox
llm-balance set-browser arc

🍪 Supported Browsers

  • Chrome (Recommended)
  • Firefox
  • Arc
  • Brave
  • Chromium

🔧 Troubleshooting

Common Issues

Q: Cookie Authentication Failed

# Ensure you're logged in to the corresponding platform
# Check global browser setting
llm-balance set-browser chrome
# Or override for single command
llm-balance cost --browser=chrome

Q: Invalid API Key

# Check if environment variables are set
echo $DEEPSEEK_API_KEY
echo $OPENAI_ADMIN_KEY

# Reset environment variables
export DEEPSEEK_API_KEY="your_valid_api_key"

Q: Specific Platform Timeout

# Test problematic platform individually
llm-balance cost --platform=deepseek

# Check network connection
curl -I https://api.deepseek.com

Configuration File Locations

  • Main configuration: ~/.llm_balance/platforms.yaml
  • Customizable via LLM_BALANCE_CONFIG_FILE environment variable

Security Notes

  • 🔒 Local Processing: All authentication information is processed locally, not uploaded to external servers
  • 🍪 Cookie Reading: Only reads necessary authentication cookies from browser, doesn't get other sensitive information
  • 🛡️ Environment Variables: API keys managed via environment variables, not written to configuration files
  • 🔄 Session Nature: Cookie authentication has时效性, requires regular platform re-login

Development Guide

Project Structure

src/llm_balance/
├── __init__.py              # Package info and version
├── cli.py                  # CLI command interface
├── balance_checker.py      # Main business logic
├── config.py              # Configuration file management
├── platform_configs.py     # Platform configuration management
├── error_handler.py       # Error handling
├── utils.py               # Utility functions
├── token_checker.py       # Token usage monitoring
├── token_formatter.py     # Token output formatting
└── platform_handlers/     # Platform handlers
    ├── __init__.py         # Handler factory
    ├── base.py            # Base handler class
    ├── aliyun.py          # Aliyun handler ✅
    ├── deepseek.py        # DeepSeek handler ✅
    ├── moonshot.py        # Moonshot handler ✅
    ├── volcengine.py      # Volcengine handler ✅
    ├── tencent.py         # Tencent handler ✅
    ├── zhipu.py           # Zhipu handler ✅
    ├── siliconflow.py     # SiliconFlow handler ✅
    ├── openai.py          # OpenAI handler ✅
    ├── anthropic.py       # Anthropic handler ✅
    ├── google.py          # Google handler ✅
    ├── foxcode.py         # FoxCode relay handler ✅
    ├── duckcoding.py      # DuckCoding relay handler ✅
    ├── _88code.py         # 88Code relay handler ✅
    ├── _88996.py          # 88996.cloud relay handler ✅
    ├── aicoding.py        # AICoding relay handler ✅
    └── yourapi.py          # YourAPI relay handler ✅

Note: Additional platform handlers (Azure OpenAI, Lingyi, MiniMax) are available in the dev branch.

Adding New Platforms

  1. Configure Platform: Add new platform configuration in ~/.llm_balance/platforms.yaml
  2. Create Handler: Inherit from BasePlatformHandler class, implement necessary authentication and parsing logic
  3. Register Handler: Add new platform in create_handler function in platform_handlers/__init__.py
  4. Test Verification: Test with llm-balance cost --platform=<new_platform>

Development Environment

# Install development dependencies
pip install -e .

# Run tests
llm-balance cost --format=json

# Debug specific platform
llm-balance cost --platform=deepseek --browser=chrome

Contributing

Issues and Pull Requests are welcome!

  1. Fork this repository
  2. Create feature branch (git checkout -b feature/new-platform)
  3. Commit changes (git commit -am 'Add new platform support')
  4. Push to branch (git push origin feature/new-platform)
  5. Create Pull Request

License

MIT License - See LICENSE file for details

About

No description, website, or topics provided.

Resources

Readme

License

MIT license
Activity

Stars

0 stars

Watchers

0 watching

Forks

0 forks
Report repository

Releases

5 tags

Packages

 
 
 

Contributors

Languages