Claude Code-Compatible Models

A comprehensive reference guide for developers looking to use Claude Code with cost-effective alternative AI providers. Includes up-to-date pricing, detailed setup instructions, and configuration examples for Anthropic, Alibaba Qwen, DeepSeek, MiniMax, Moonshot AI (Kimi K2), and Zhipu GLM models.

Provider	Model	Input $/1M tok	Output $/1M tok	Notes
Anthropic	Haiku 4.5	$1.00	$5.00	Official API pricing.
Anthropic	Opus 4.1	$15.00	$75.00	Official API pricing.
Anthropic	Sonnet 4.5	$3.00	$15.00	Official API pricing.
Alibaba	Qwen3 Coder	$0.22	$0.95	Uses Anthropic-compatible endpoint.
DeepSeek	deepseek-chat	$0.28	$0.42	Context caching enabled by default.
MiniMax	MiniMax-M2	$0.30	$1.20	Has Coding Plan from $10/mo.
Moonshot AI	Kimi K2	$0.14	$2.49	Automatic prompt caching (beta).
Zhipu (ZAI)	GLM 4.6	$0.60	$2.20	Has GLM Coding Plan from $3/mo.

---

Provider Setup Guides

Anthropic (Official)

Installation:

npm install -g @anthropic-ai/claude-code
# or on Unix systems:
curl -fsSL https://claude.ai/install.sh | bash

Authentication:

There are two main options:

1. Claude Console (OAuth): Requires active billing at console.anthropic.com. A "Claude Code" workspace is automatically created for usage tracking.

2. Claude.ai Subscription: Subscribe to Claude Pro or Max plan and log in with your Claude.ai account.

Enterprise Options:

• Configure Claude Code to use Amazon Bedrock or Google Vertex AI with your existing cloud infrastructure.

Documentation: https://docs.claude.com/en/docs/claude-code/setup

---

Alibaba - Qwen

Getting API Key:

1. Visit Alibaba Cloud Model Studio
2. Create an account and generate a DashScope API key

Configuration:

Edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://dashscope-intl.aliyuncs.com/apps/anthropic",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_DASHSCOPE_API_KEY",
    "ANTHROPIC_MODEL": "qwen-plus",
    "ANTHROPIC_SMALL_FAST_MODEL": "qwen-flash"
  }
}

Recommended Model Combinations:

• Main Model (ANTHROPIC_MODEL): qwen-max, qwen-plus, or Qwen-Coder series for complex tasks
• Fast Model (ANTHROPIC_SMALL_FAST_MODEL): qwen-flash or qwen-turbo for simple jobs

Launch:

cd your-project-directory
claude

Documentation: https://www.alibabacloud.com/help/en/model-studio/claude-code

---

DeepSeek

Getting API Key: Visit DeepSeek Platform to obtain your API key.

Configuration:

Edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.deepseek.com/anthropic",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_DEEPSEEK_API_KEY",
    "API_TIMEOUT_MS": "600000",
    "ANTHROPIC_MODEL": "deepseek-chat",
    "ANTHROPIC_SMALL_FAST_MODEL": "deepseek-chat",
    "CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC": "1"
  }
}

Using DeepSeek Reasoner (R1): Replace deepseek-chat with deepseek-reasoner in the configuration above.

Launch:

cd your-project-directory
claude

Troubleshooting:

• Invalid token error: Verify ANTHROPIC_AUTH_TOKEN is set to your real DeepSeek API key
• No response: Confirm ANTHROPIC_BASE_URL is correct: https://api.deepseek.com/anthropic

Documentation: https://api-docs.deepseek.com/guides/anthropic_api

---

MiniMax

Getting API Key:

1. Visit MiniMax Developer Platform
2. Navigate to the interface key section
3. Create a new secret key with a project name
4. Copy and save the key securely (shown only once)

Configuration:

Edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.minimax.io/anthropic",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_MINIMAX_API_KEY",
    "API_TIMEOUT_MS": "3000000",
    "ANTHROPIC_MODEL": "MiniMax-M2",
    "ANTHROPIC_SMALL_FAST_MODEL": "MiniMax-M2",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "MiniMax-M2",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "MiniMax-M2",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "MiniMax-M2",
    "CLAUDE_CODE_DISABLE_NONESSENTIAL_TRAFFIC": "1"
  }
}

Launch:

cd your-project-directory
claude

Documentation: https://platform.minimax.io/docs/guides/text-ai-coding-tools

---

Moonshot AI - Kimi K2

Getting API Key:

1. Visit Moonshot Platform
2. Create an account and generate an API key

Configuration:

Edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.moonshot.ai/anthropic",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_MOONSHOT_API_KEY",
    "ANTHROPIC_MODEL": "kimi-k2-turbo-preview",
    "ANTHROPIC_SMALL_FAST_MODEL": "kimi-k2-turbo-preview"
  }
}

Launch:

cd your-project-directory
claude

Features:

• 256K context length for handling complex codebases
• Seamless Claude Code compatibility with agentic tool use
• Automatic prompt caching (currently in 3-month public beta)

Documentation: https://platform.moonshot.ai/docs/guide/agent-support

---

Zhipu (ZAI) - GLM 4.6

Getting API Key:

1. Visit Z.AI Platform (Mainland China) or api.z.ai (International)
2. Create an account and generate an API key

Configuration:

Edit ~/.claude/settings.json:

{
  "env": {
    "ANTHROPIC_BASE_URL": "https://api.z.ai/api/anthropic",
    "ANTHROPIC_AUTH_TOKEN": "YOUR_ZAI_API_KEY",
    "ANTHROPIC_DEFAULT_HAIKU_MODEL": "glm-4.5-air",
    "ANTHROPIC_DEFAULT_SONNET_MODEL": "glm-4.6",
    "ANTHROPIC_DEFAULT_OPUS_MODEL": "glm-4.6"
  }
}

For Mainland China: Replace base URL with: https://open.bigmodel.cn/api/anthropic

Launch:

cd your-project-directory
claude

Documentation: https://docs.z.ai/scenario-example/develop-tools/claude

---

💰 Coding Plans Overview

Several providers offer subscription-based coding plans that provide generous limits for development work, often at 3-10x better value than pay-per-token pricing. These are perfect for intensive coding sessions or professional development work.

MiniMax M2 - Coding Plan ⭐

Plans:

• Starter: $10/mo (100 prompts/5h) | $100/yr
• Plus: $20/mo (300 prompts/5h) | $200/yr
• Max: $50/mo (1000 prompts/5h) | $500/yr

Benefits:

• 1 prompt ≈ 15 model requests
• 1/10th the cost of comparable plans (Claude, etc.)
• No unpredictable bills - fixed subscription
• Powered by MiniMax-M2 (230B MoE, 10B active)
• 204K context window

Subscribe: https://platform.minimax.io/subscribe/coding-plan

---

GLM 4.6 - GLM Coding Plan ⭐

Plans:

• Lite: $3/mo (~120 prompts/5h)
• Pro: $15/mo (~600 prompts/5h)
• Max: Price varies (~2400 prompts/5h)

Benefits:

• 3x more usage than comparable plans
• 1 prompt = ~15-20 model calls
• Over 55 tokens/second generation speed
• No network restrictions or account bans
• Vision & Web Search (Pro+)
• Compatible with Claude Code, Cline, Cursor, etc.

Features:

• Text generation, Vision, Video generation
• Model mapping: Opus/Sonnet/Haiku all default to GLM models
• Singapore-based services with privacy protection

Subscribe: https://z.ai/subscribe

---

General Tips

1. Version Check: Ensure Claude Code is up to date with claude --version and update with claude update

2. Timeout Settings: Some providers require higher timeout values (especially for complex reasoning tasks). The API_TIMEOUT_MS setting helps prevent premature timeouts.

3. Model Selection: Most providers support automatic model selection based on task complexity using ANTHROPIC_MODEL (main) and ANTHROPIC_SMALL_FAST_MODEL (quick tasks).

4. Regional Considerations: China-based providers often have separate endpoints for mainland China vs. international users.

5. Testing Configuration: After setup, run claude in a test directory and verify the model responds correctly.

6. Cost Optimization:

   • Use cache-enabled providers (Anthropic, DeepSeek, Kimi) for repetitive tasks
   • Select appropriate model sizes for the task complexity
   • Consider batch processing for non-urgent workloads