For OpenClaw v2026.2 | This article assumes you have completed basic configuration and need to optimize model selection or cost control.

TL;DR: OpenClaw supports multiple models: Claude Opus/Sonnet/Haiku, GPT-4o, Gemini, local models (Ollama). Configure "agent": {"model": "anthropic/claude-opus-4-6"} to switch. Failover automatically switches to fallback: "fallback": [{"model": "claude-sonnet-4", "condition": "rate_limit"}]. Recommended: Claude Opus for complex tasks, Haiku for high-frequency simple tasks.

Supported Model Providers

Provider Overview

Model Capability Comparison

Basic Model Configuration

Anthropic Claude

{
  "agent": {
    "model": "anthropic/claude-opus-4-6",
    "thinkingLevel": "high"
  },
  "models": {
    "anthropic": {
      "apiKey": "${ANTHROPIC_API_KEY}",
      "baseUrl": "https://api.anthropic.com"
    }
  }
}

Claude Model Selection Guide:

OpenAI GPT

{
  "agent": {
    "model": "openai/gpt-4o"
  },
  "models": {
    "openai": {
      "apiKey": "${OPENAI_API_KEY}",
      "baseUrl": "https://api.openai.com/v1",
      "organization": "org-xxx"
    }
  }
}

GPT Model Selection Guide:

Google Gemini

{
  "agent": {
    "model": "google/gemini-2.0-flash"
  },
  "models": {
    "google": {
      "apiKey": "${GOOGLE_API_KEY}",
      "baseUrl": "https://generativelanguage.googleapis.com/v1beta"
    }
  }
}

Local Models (Ollama)

{
  "agent": {
    "model": "ollama/llama3.1:70b"
  },
  "models": {
    "ollama": {
      "baseUrl": "http://localhost:11434"
    }
  }
}

Ollama Installation and Startup:

# Install Ollama
curl -fsSL https://ollama.com/install.sh | sh

# Download models
ollama pull llama3.1:70b
ollama pull codellama:34b

# Start service
ollama serve

Third-Party API Proxy

{
  "agent": {
    "model": "anthropic/claude-opus-4-6"
  },
  "models": {
    "anthropic": {
      "apiKey": "${PROVIDER_API_KEY}",
      "baseUrl": "https://api.your-provider.com/v1"
    }
  }
}

Model Format Reference

Model Identifier Format

{provider}/{model-name}

Custom Model Aliases

{
  "models": {
    "aliases": {
      "smart": "anthropic/claude-opus-4-6",
      "fast": "anthropic/claude-sonnet-4",
      "cheap": "anthropic/claude-haiku-3.5",
      "local": "ollama/llama3.1:70b"
    }
  },
  "agent": {
    "model": "smart"
  }
}

Using aliases:

# Switch in chat
/model fast

# Or use full name
/model anthropic/claude-sonnet-4

Failover Strategy

What is Failover?

When the primary model is unavailable (e.g., rate limit reached, service outage), automatically switch to a fallback model.

flowchart TB
    REQ[Request arrives] --> PRIMARY{Primary available?}
    PRIMARY -->|Yes| EXEC1[Use primary]
    PRIMARY -->|No| CHECK{Failover condition}
    CHECK -->|Rate limit| SECONDARY1[Fallback 1]
    CHECK -->|Service outage| SECONDARY2[Fallback 2]
    CHECK -->|Timeout| SECONDARY3[Fallback 3]
    SECONDARY1 --> EXEC2[Execute request]
    SECONDARY2 --> EXEC2
    SECONDARY3 --> EXEC2
    EXEC1 --> RES[Return result]
    EXEC2 --> RES

Configuring Failover

{
  "agent": {
    "model": "anthropic/claude-opus-4-6"
  },
  "models": {
    "fallback": [
      {
        "model": "anthropic/claude-sonnet-4",
        "condition": "rate_limit"
      },
      {
        "model": "openai/gpt-4o",
        "condition": "error"
      },
      {
        "model": "ollama/llama3.1:70b",
        "condition": "timeout"
      }
    ]
  }
}

Failover Conditions

Advanced Failover Configuration

{
  "models": {
    "fallback": [
      {
        "model": "anthropic/claude-sonnet-4",
        "condition": "rate_limit",
        "maxRetries": 3,
        "retryDelay": 1000
      },
      {
        "model": "openai/gpt-4o",
        "condition": "error",
        "maxRetries": 2
      },
      {
        "model": "ollama/llama3.1:70b",
        "condition": "always"
      }
    ],
    "fallbackPolicy": {
      "retryPrimaryAfter": 60000,
      "logFallbacks": true,
      "notifyUser": true
    }
  }
}

Scene-Based Model Switching

Switch by Session Type

Use different models for different session types:

{
  "agents": {
    "defaults": {
      "model": "anthropic/claude-sonnet-4"
    },
    "byChannel": {
      "telegram": {
        "model": "anthropic/claude-haiku-3.5"
      },
      "discord": {
        "model": "anthropic/claude-sonnet-4"
      }
    },
    "bySessionType": {
      "main": {
        "model": "anthropic/claude-opus-4-6"
      },
      "group": {
        "model": "anthropic/claude-haiku-3.5"
      }
    }
  }
}

Switch by Task Type

{
  "agents": {
    "taskModels": {
      "code": "anthropic/claude-opus-4-6",
      "reasoning": "anthropic/claude-opus-4-6",
      "chat": "anthropic/claude-sonnet-4",
      "simple": "anthropic/claude-haiku-3.5"
    }
  }
}

Dynamic Model Selection

Switch models at runtime:

# CLI switch
openclaw agent --model anthropic/claude-opus-4-6 --message "Complex task"

# Switch in chat
/model anthropic/claude-opus-4-6
Now using Claude Opus 4.6, suitable for complex reasoning

/model anthropic/claude-haiku-3.5
Switched to Haiku 3.5, fast response for simple tasks

Cost Optimization Strategies

Model Cost Comparison

Cost Optimization Configuration

{
  "agents": {
    "defaults": {
      "model": "anthropic/claude-sonnet-4",
      "costOptimization": {
        "enabled": true,
        "rules": [
          {
            "condition": "context_length > 50000",
            "model": "anthropic/claude-sonnet-4"
          },
          {
            "condition": "simple_query",
            "model": "anthropic/claude-haiku-3.5"
          },
          {
            "condition": "code_generation",
            "model": "anthropic/claude-opus-4-6"
          }
        ]
      }
    }
  }
}

Token Usage Monitoring

# View current session usage
/status

# Output example:
# Session: main
# Model: anthropic/claude-opus-4-6
# Tokens: 12,345 input / 2,456 output
# Cost: $0.21

# Enable usage display
/usage full

Usage Display Configuration

{
  "usageTracking": {
    "enabled": true,
    "showInResponse": true,
    "trackBySession": true,
    "trackByChannel": true,
    "alertThreshold": 10.00
  }
}

Thinking Level Configuration

Thinking Level Reference

Configuring Thinking Level

{
  "agent": {
    "model": "anthropic/claude-opus-4-6",
    "thinkingLevel": "high"
  }
}

Dynamic Switching

# Switch in chat
/think high
Now using high thinking mode, suitable for complex reasoning

/think minimal
Switched to minimal thinking, fast response

Local Model Configuration Details

Ollama Configuration

{
  "agent": {
    "model": "ollama/llama3.1:70b"
  },
  "models": {
    "ollama": {
      "baseUrl": "http://localhost:11434",
      "options": {
        "temperature": 0.7,
        "num_ctx": 32768,
        "num_predict": 4096
      }
    }
  }
}

Recommended Local Models

Hardware Recommendations

Multi-Provider Load Balancing

Round-Robin Strategy

{
  "models": {
    "loadBalancing": {
      "strategy": "round_robin",
      "providers": [
        {
          "model": "anthropic/claude-sonnet-4",
          "weight": 1
        },
        {
          "model": "openai/gpt-4o",
          "weight": 1
        }
      ]
    }
  }
}

Weighted Strategy

{
  "models": {
    "loadBalancing": {
      "strategy": "weighted",
      "providers": [
        {
          "model": "anthropic/claude-sonnet-4",
          "weight": 70
        },
        {
          "model": "openai/gpt-4o",
          "weight": 20
        },
        {
          "model": "ollama/llama3.1:70b",
          "weight": 10
        }
      ]
    }
  }
}

Cost-Based Strategy

{
  "models": {
    "loadBalancing": {
      "strategy": "cost_optimized",
      "dailyBudget": 10.00,
      "providers": [
        {
          "model": "anthropic/claude-opus-4-6",
          "maxBudget": 5.00,
          "useFor": ["complex", "reasoning"]
        },
        {
          "model": "anthropic/claude-sonnet-4",
          "maxBudget": 3.00,
          "useFor": ["chat", "code"]
        },
        {
          "model": "anthropic/claude-haiku-3.5",
          "maxBudget": 2.00,
          "useFor": ["simple", "high_frequency"]
        }
      ]
    }
  }
}

Model Configuration Best Practices

1. Layer by Scenario

{
  "agents": {
    "defaults": {
      "model": "anthropic/claude-sonnet-4",
      "thinkingLevel": "medium"
    },
    "byUseCase": {
      "coding": {
        "model": "anthropic/claude-opus-4-6",
        "thinkingLevel": "high"
      },
      "chat": {
        "model": "anthropic/claude-haiku-3.5",
        "thinkingLevel": "minimal"
      },
      "research": {
        "model": "anthropic/claude-opus-4-6",
        "thinkingLevel": "xhigh"
      }
    }
  }
}

2. Set Reasonable Budget

{
  "budget": {
    "daily": 20.00,
    "weekly": 100.00,
    "monthly": 400.00,
    "alertThreshold": 0.8,
    "hardLimit": true
  }
}

3. Monitor and Analyze

{
  "analytics": {
    "trackUsage": true,
    "trackCosts": true,
    "trackPerformance": true,
    "exportPath": "~/.openclaw/analytics",
    "retentionDays": 30
  }
}

4. Test and Validate

# Test model connection
openclaw doctor --check models

# Test specific model
openclaw agent --model anthropic/claude-opus-4-6 --message "Test message"

# Compare model responses
openclaw benchmark --models "anthropic/claude-opus-4-6,openai/gpt-4o" --prompt "Complex question"

Troubleshooting

Model Connection Failed

# Check API Key
openclaw config get models.anthropic.apiKey

# Test connection
curl -X POST https://api.anthropic.com/v1/messages \
  -H "x-api-key: $ANTHROPIC_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "claude-sonnet-4", "max_tokens": 10, "messages": [{"role": "user", "content": "hi"}]}'

# Check network
ping api.anthropic.com

Rate Limiting

# View usage
openclaw usage --today

# Output example:
# Date: 2026-02-26
# Total tokens: 123,456
# Cost: $1.23
# Requests: 89

# Configure rate limit handling
openclaw config set models.anthropic.rateLimitHandling auto_retry

Local Model Issues

# Check Ollama service
curl http://localhost:11434/api/tags

# Check if model is loaded
ollama list

# Manually load model
ollama run llama3.1:70b

# Check GPU usage
nvidia-smi

Summary

Multi-model configuration is one of OpenClaw’s powerful features:

• Flexible switching: Choose the right model for each task
• Failover protection: Automatically handle model unavailability
• Cost optimization: Allocate budget wisely, avoid waste
• Local support: Use local models for privacy-sensitive scenarios

With proper model configuration, you can find the best balance between performance, cost, and privacy.

Article Summary:

• Mastered configuration for multiple model providers
• Understood Failover strategy and configuration
• Learned scene-based dynamic model switching
• Explored cost optimization and monitoring
• Mastered local model configuration and usage

Changelog:

• 2026-02-26: Initial release, based on OpenClaw v2026.2

Series Navigation:

• ← Previous: Gateway Core Concepts Explained
• → Next: Skills System Explained