Home / Integrations / Model Providers

☁️ Model Providers

Code Puppy supports multiple AI providers and models. Use your favorite models or mix and match!

Supported Providers

🟢 OpenAI

GPT-4, GPT-4o, GPT-3.5-turbo, and more.

Env: OPENAI_API_KEY

🔵 Anthropic

Claude 3.5 Sonnet, Claude 3 Opus, Haiku.

Env: ANTHROPIC_API_KEY

🔴 Google

Gemini Pro, Gemini Ultra, and more.

Env: GOOGLE_API_KEY

🟡 Cerebras

Ultra-fast inference with Cerebras chips.

Env: CEREBRAS_API_KEY

🟠 Mistral

Mistral Large, Medium, and open models.

Env: MISTRAL_API_KEY

🟪 Groq

Fast LLM inference with LPU chips.

Env: GROQ_API_KEY

Setting Up API Keys

Option 1: Using /set Command (Recommended)

Set API keys directly in Code Puppy using lowercase key names. Keys are saved to puppy.cfg and persist across sessions:

/set openai_api_key sk-your-openai-key
/set anthropic_api_key sk-ant-your-anthropic-key
/set google_api_key your-google-key
/set cerebras_api_key your-cerebras-key
/set mistral_api_key your-mistral-key
/set groq_api_key your-groq-key

✅ No Environment Variables Needed

Keys set with /set are stored in ~/.code_puppy/puppy.cfg and loaded automatically. Much easier than managing environment variables!

Option 2: Environment Variables

Alternatively, use environment variables (add to ~/.bashrc or ~/.zshrc):

export OPENAI_API_KEY="sk-your-openai-key"
export ANTHROPIC_API_KEY="sk-ant-your-anthropic-key"
export GOOGLE_API_KEY="your-google-key"
export CEREBRAS_API_KEY="your-cerebras-key"
export MISTRAL_API_KEY="your-mistral-key"
export GROQ_API_KEY="your-groq-key"

Switching Models

# Switch model on the fly
/model gpt-5.1
/model claude-4-5-sonnet
/model Gemini-3

# List available models
/model

# Add models from catalog
/add_model

Popular Models

Provider	Model	Best For
OpenAI	`gpt-5.1`	General purpose, fast
OpenAI	`gpt-5.1-codex-api`	Complex reasoning
Anthropic	`claude-4-5-sonnet`	Code generation, analysis
Anthropic	`claude-4-5-opus`	Most capable Claude
Anthropic	`claude-4-5-haiku`	Fast, cheap
Google	`Gemini-3`	General purpose
Cerebras	`Cerebras-GLM-4.6`	Very fast inference
Groq	`synthetic-GLM-4.6`	Fast open models

Round-Robin Mode

Distribute requests across multiple models for load balancing or cost optimization:

# In puppy.cfg
[round_robin]
enabled = true
models = gpt-5.1, claude-4-5-sonnet, Gemini-3

Adding Custom Models

Add custom models via ~/.code_puppy/extra_models.json:

{
  "models": [
    {
      "name": "my-local-llama",
      "provider": "openai",
      "model_id": "Cerebras-GLM-4.6",
      "api_base": "http://localhost:8000/v1"
    },
    {
      "name": "azure-gpt4",
      "provider": "azure",
      "model_id": "gpt-5.1",
      "api_base": "https://my-deployment.openai.azure.com",
      "api_version": "2024-02-15-preview"
    }
  ]
}

Per-Model Settings

Configure settings for specific models:

/model_settings

# Adjust temperature, max tokens, etc.
# Settings are saved per-model

Model Pinning

Pin specific models to agents:

# Pin claude to the code-reviewer agent
/agent code-reviewer
/pin_model claude-4-5-sonnet

# Now code-reviewer always uses Claude
# Other agents use your default model