AI Providers

​

Supported Providers

• Anthropic - Claude models (3.5 Sonnet, 3.7 Sonnet, Opus, Haiku)
• OpenAI - GPT models (GPT-4o, GPT-4o-mini, o1, o3-mini)
• Google AI - Gemini models (Gemini 2.0, Gemini 1.5 Pro)
• xAI - Grok models
• Zed Cloud - Proxied access to various models
• Ollama - Local open-source models
• LM Studio - Local model hosting
• DeepSeek - DeepSeek models
• Mistral - Mistral models
• OpenRouter - Access to multiple providers
• Vercel AI - Vercel AI SDK compatible models
• AWS Bedrock - Amazon Bedrock models
• Custom OpenAI-compatible APIs

​

Quick Start: Zed Cloud (Easiest)

1. Sign in to Zed with your GitHub account
2. Open Settings (Cmd+,) → Language Models
3. Zed Cloud should be automatically configured
4. Select a model and start using AI features

• Claude 3.5 Sonnet
• Claude 3.5 Haiku
• GPT-4o
• GPT-4o-mini
• Gemini 1.5 Pro

• Rate limits based on your Zed subscription
• Requires internet connection
• Limited model selection

​

Anthropic (Claude)

​

Getting an API Key

1. Sign up at console.anthropic.com
2. Navigate to API Keys
3. Create a new API key
4. Copy the key (starts with sk-ant-)

​

Configuration in Zed

1. Open Settings → Language Models
2. Under “Anthropic”, click “Authenticate”
3. Paste your API key
4. Select a default model

​

Settings Example

{
  "language_models": {
    "anthropic": {
      "api_url": "https://api.anthropic.com",
      "available_models": [
        {
          "name": "claude-3-5-sonnet-20241022",
          "display_name": "Claude 3.5 Sonnet",
          "max_tokens": 200000,
          "max_output_tokens": 8192,
          "cache_configuration": {
            "max_cache_anchors": 10,
            "should_speculate": true,
            "min_total_token": 2048
          }
        },
        {
          "name": "claude-3-7-sonnet-20250219",
          "display_name": "Claude 3.7 Sonnet (Extended Thinking)",
          "max_tokens": 200000,
          "max_output_tokens": 8192,
          "mode": {
            "type": "thinking",
            "budget_tokens": 10000
          }
        }
      ]
    }
  }
}

• claude-3-5-sonnet-20241022 - Best balance of intelligence and speed
• claude-3-7-sonnet-20250219 - Extended thinking for complex tasks
• claude-3-5-haiku-20241022 - Fast and economical

​

Prompt Caching

{
  "cache_configuration": {
    "max_cache_anchors": 10,      // Number of cache breakpoints
    "should_speculate": true,       // Speculative caching
    "min_total_token": 2048         // Minimum tokens to enable caching
  }
}

​

OpenAI

​

Getting an API Key

1. Sign up at platform.openai.com
2. Navigate to API Keys
3. Create a new secret key
4. Copy the key (starts with sk-)

​

Configuration in Zed

1. Open Settings → Language Models
2. Under “OpenAI”, click “Authenticate”
3. Paste your API key
4. Select a default model

​

Settings Example

{
  "language_models": {
    "openai": {
      "api_url": "https://api.openai.com/v1",
      "available_models": [
        {
          "name": "gpt-4o",
          "display_name": "GPT-4o",
          "max_tokens": 128000,
          "max_output_tokens": 16384
        },
        {
          "name": "gpt-4o-mini",
          "display_name": "GPT-4o Mini",
          "max_tokens": 128000,
          "max_output_tokens": 16384
        },
        {
          "name": "o1",
          "display_name": "o1",
          "max_tokens": 200000,
          "max_completion_tokens": 100000,
          "reasoning_effort": "medium"
        }
      ]
    }
  }
}

• gpt-4o - Latest and most capable
• gpt-4o-mini - Fast and cost-effective
• o1 - Reasoning-focused for complex problems

​

Google AI (Gemini)

​

Getting an API Key

1. Sign up at aistudio.google.com
2. Click “Get API key”
3. Copy the generated key

​

Configuration in Zed

1. Open Settings → Language Models
2. Under “Google AI”, click “Authenticate”
3. Paste your API key

​

Settings Example

{
  "language_models": {
    "google": {
      "api_url": "https://generativelanguage.googleapis.com",
      "available_models": [
        {
          "name": "gemini-2.0-flash-exp",
          "display_name": "Gemini 2.0 Flash",
          "max_tokens": 1048576
        },
        {
          "name": "gemini-1.5-pro-002",
          "display_name": "Gemini 1.5 Pro",
          "max_tokens": 2097152
        }
      ]
    }
  }
}

• gemini-2.0-flash-exp - Fast multimodal model
• gemini-1.5-pro-002 - Extremely large context window (2M tokens)

​

Ollama (Local Models)

​

Installation

1. Download Ollama from ollama.ai
2. Install the application
3. Pull a model: ollama pull qwen2.5-coder:32b
4. Start Ollama (usually starts automatically)

​

Configuration in Zed

{
  "language_models": {
    "ollama": {
      "api_url": "http://localhost:11434",
      "auto_discover": true,
      "available_models": [
        {
          "name": "qwen2.5-coder:32b",
          "display_name": "Qwen 2.5 Coder 32B",
          "max_tokens": 32768,
          "supports_tools": true,
          "supports_images": false,
          "keep_alive": "10m"
        },
        {
          "name": "deepseek-r1:32b",
          "display_name": "DeepSeek R1 32B",
          "max_tokens": 65536,
          "supports_tools": true,
          "supports_thinking": true
        }
      ]
    }
  }
}

• qwen2.5-coder:32b - Excellent for code generation
• deepseek-r1:32b - Strong reasoning capabilities
• llama3.2:3b - Fast and lightweight

• Complete privacy (runs locally)
• No API costs
• No rate limits
• Works offline

• Powerful GPU recommended (32B models need ~20GB VRAM)
• Or use CPU with sufficient RAM (slower)

​

Custom OpenAI-Compatible APIs

{
  "language_models": {
    "openai_compatible": {
      "my_local_llm": {
        "api_url": "http://localhost:8080/v1",
        "available_models": [
          {
            "name": "my-model",
            "display_name": "My Custom Model",
            "max_tokens": 8192,
            "capabilities": {
              "tools": true,
              "images": false,
              "parallel_tool_calls": true
            }
          }
        ]
      }
    }
  }
}

• LM Studio
• Text Generation Web UI
• vLLM
• LocalAI
• Any OpenAI-compatible inference server

​

AWS Bedrock

​

Prerequisites

1. AWS account with Bedrock access
2. AWS CLI installed and configured
3. Model access enabled in Bedrock console

​

Configuration

{
  "language_models": {
    "bedrock": {
      "region": "us-east-1",
      "profile": "default",
      "authentication_method": "named_profile",
      "available_models": [
        {
          "name": "anthropic.claude-3-5-sonnet-20241022-v2:0",
          "display_name": "Claude 3.5 Sonnet (Bedrock)",
          "max_tokens": 200000
        }
      ]
    }
  }
}

• named_profile - Use AWS CLI profile
• sso - AWS SSO
• api_key - Access key and secret
• default - Environment variables or instance role

​

OpenRouter

​

Getting an API Key

1. Sign up at openrouter.ai
2. Navigate to Keys
3. Create a new key

​

Configuration

{
  "language_models": {
    "open_router": {
      "api_url": "https://openrouter.ai/api/v1",
      "available_models": [
        {
          "name": "anthropic/claude-3.5-sonnet",
          "display_name": "Claude 3.5 Sonnet (via OpenRouter)",
          "max_tokens": 200000
        },
        {
          "name": "openai/gpt-4o",
          "display_name": "GPT-4o (via OpenRouter)",
          "max_tokens": 128000
        }
      ]
    }
  }
}

• Single API key for multiple providers
• Automatic fallback between providers
• Pay-as-you-go pricing

​

Comparing Providers

​

Troubleshooting

​

”Missing API key” error

1. Check Settings → Language Models
2. Click “Authenticate” for your provider
3. Ensure the key is correct (no extra spaces)
4. Try generating a new API key

​

”Rate limit exceeded”

• Wait before retrying
• Check your provider’s usage limits
• Consider upgrading your account
• Use a different provider

​

”Model not found”

• Verify the model name exactly matches the provider’s API
• Check that you have access to the model
• For Bedrock, ensure model access is enabled

​

Ollama not connecting

1. Ensure Ollama is running: ollama list
2. Check the API URL: http://localhost:11434
3. Try pulling the model: ollama pull <model-name>
4. Check firewall settings

​

Slow responses

• For local models: upgrade hardware or use smaller models
• For API providers: check internet connection
• Try a different model (e.g., mini variants)

​

Best Practices

1. Start with Zed Cloud for quick setup
2. Keep API keys secure - never commit them to git
3. Use environment variables for shared configurations
4. Monitor costs on paid providers
5. Try local models for privacy-sensitive projects
6. Use different providers for different tasks
7. Cache aggressively with Anthropic to reduce costs

​

Next Steps

• Configure AI settings
• Learn to use the Agent Panel
• Master the Inline Assistant