Configuration

There are various aspects about the Agent Panel that you can customize. All of them can be seen by either visiting the Configuring Zed page or by running the zed: open default settings action and searching for "agent".

Alternatively, you can also visit the panel's Settings view by running the agent: open configuration action or going to the top-right menu and hitting "Settings".

LLM Providers

Zed supports multiple large language model providers. Here's an overview of the supported providers and tool call support:

Provider	Tool Use Supported
Amazon Bedrock	Depends on the model
Anthropic	✅
DeepSeek	✅
GitHub Copilot Chat	For some models (link)
Google AI	✅
LM Studio	✅
Mistral	✅
Ollama	✅
OpenAI	✅
OpenAI API Compatible	🚫
OpenRouter	✅

Use Your Own Keys

While Zed offers hosted versions of models through our various plans, we're always happy to support users wanting to supply their own API keys. Below, you can learn how to do that for each provider.

Amazon Bedrock

To use Amazon Bedrock's models, an AWS authentication is required. Ensure your credentials have the following permissions set up:

bedrock:InvokeModelWithResponseStream
bedrock:InvokeModel
bedrock:ConverseStream

Your IAM policy should look similar to:

{
  "Version": "2012-10-17",
  "Statement": [
    {
      "Effect": "Allow",
      "Action": [
        "bedrock:InvokeModel",
        "bedrock:InvokeModelWithResponseStream",
        "bedrock:ConverseStream"
      ],
      "Resource": "*"
    }
  ]
}

With that done, choose one of the two authentication methods:

Authentication via Named Profile (Recommended)

Ensure you have the AWS CLI installed and configured with a named profile

Open your settings.json (zed: open settings) and include the bedrock key under language_models with the following settings:

{
  "language_models": {
    "bedrock": {
      "authentication_method": "named_profile",
      "region": "your-aws-region",
      "profile": "your-profile-name"
    }
  }
}

Authentication via Static Credentials

While it's possible to configure through the Agent Panel settings UI by entering your AWS access key and secret directly, we recommend using named profiles instead for better security practices. To do this:

Create an IAM User that you can assume in the IAM Console.
Create security credentials for that User, save them and keep them secure.
Open the Agent Configuration with (agent: open configuration) and go to the Amazon Bedrock section
Copy the credentials from Step 2 into the respective Access Key ID, Secret Access Key, and Region fields.

Cross-Region Inference

The Zed implementation of Amazon Bedrock uses Cross-Region inference for all the models and region combinations that support it. With Cross-Region inference, you can distribute traffic across multiple AWS Regions, enabling higher throughput.

For example, if you use Claude Sonnet 3.7 Thinking from us-east-1, it may be processed across the US regions, namely: us-east-1, us-east-2, or us-west-2. Cross-Region inference requests are kept within the AWS Regions that are part of the geography where the data originally resides. For example, a request made within the US is kept within the AWS Regions in the US.

Although the data remains stored only in the source Region, your input prompts and output results might move outside of your source Region during cross-Region inference. All data will be transmitted encrypted across Amazon's secure network.

We will support Cross-Region inference for each of the models on a best-effort basis, please refer to the Cross-Region Inference method Code.

For the most up-to-date supported regions and models, refer to the Supported Models and Regions for Cross Region inference.

Anthropic

You can use Anthropic models by choosing it via the model dropdown in the Agent Panel.

Sign up for Anthropic and create an API key
Make sure that your Anthropic account has credits
Open the settings view (agent: open configuration) and go to the Anthropic section
Enter your Anthropic API key

Even if you pay for Claude Pro, you will still have to pay for additional credits to use it via the API.

Zed will also use the ANTHROPIC_API_KEY environment variable if it's defined.

Custom Models

You can add custom models to the Anthropic provider by adding the following to your Zed settings.json:

{
  "language_models": {
    "anthropic": {
      "available_models": [
        {
          "name": "claude-3-5-sonnet-20240620",
          "display_name": "Sonnet 2024-June",
          "max_tokens": 128000,
          "max_output_tokens": 2560,
          "cache_configuration": {
            "max_cache_anchors": 10,
            "min_total_token": 10000,
            "should_speculate": false
          },
          "tool_override": "some-model-that-supports-toolcalling"
        }
      ]
    }
  }
}

Custom models will be listed in the model dropdown in the Agent Panel.

You can configure a model to use extended thinking (if it supports it) by changing the mode in your model's configuration to thinking, for example:

{
  "name": "claude-sonnet-4-latest",
  "display_name": "claude-sonnet-4-thinking",
  "max_tokens": 200000,
  "mode": {
    "type": "thinking",
    "budget_tokens": 4_096
  }
}

DeepSeek

Visit the DeepSeek platform and create an API key
Open the settings view (agent: open configuration) and go to the DeepSeek section
Enter your DeepSeek API key

The DeepSeek API key will be saved in your keychain.

Zed will also use the DEEPSEEK_API_KEY environment variable if it's defined.

Custom Models

The Zed agent comes pre-configured to use the latest version for common models (DeepSeek Chat, DeepSeek Reasoner). If you wish to use alternate models or customize the API endpoint, you can do so by adding the following to your Zed settings.json:

{
  "language_models": {
    "deepseek": {
      "api_url": "https://api.deepseek.com",
      "available_models": [
        {
          "name": "deepseek-chat",
          "display_name": "DeepSeek Chat",
          "max_tokens": 64000
        },
        {
          "name": "deepseek-reasoner",
          "display_name": "DeepSeek Reasoner",
          "max_tokens": 64000,
          "max_output_tokens": 4096
        }
      ]
    }
  }
}

Custom models will be listed in the model dropdown in the Agent Panel. You can also modify the api_url to use a custom endpoint if needed.

GitHub Copilot Chat

You can use GitHub Copilot Chat with the Zed agent by choosing it via the model dropdown in the Agent Panel.

Open the settings view (agent: open configuration) and go to the GitHub Copilot Chat section
Click on Sign in to use GitHub Copilot, follow the steps shown in the modal.

Alternatively, you can provide an OAuth token via the GH_COPILOT_TOKEN environment variable.

Google AI

You can use Gemini models with the Zed agent by choosing it via the model dropdown in the Agent Panel.

Go to the Google AI Studio site and create an API key.
Open the settings view (agent: open configuration) and go to the Google AI section
Enter your Google AI API key and press enter.

The Google AI API key will be saved in your keychain.

Zed will also use the GOOGLE_AI_API_KEY environment variable if it's defined.

Custom Models

By default, Zed will use stable versions of models, but you can use specific versions of models, including experimental models. You can configure a model to use thinking mode (if it supports it) by adding a mode configuration to your model. This is useful for controlling reasoning token usage and response speed. If not specified, Gemini will automatically choose the thinking budget.

Here is an example of a custom Google AI model you could add to your Zed settings.json:

{
  "language_models": {
    "google": {
      "available_models": [
        {
          "name": "gemini-2.5-flash-preview-05-20",
          "display_name": "Gemini 2.5 Flash (Thinking)",
          "max_tokens": 1000000,
          "mode": {
            "type": "thinking",
            "budget_tokens": 24000
          }
        }
      ]
    }
  }
}

Custom models will be listed in the model dropdown in the Agent Panel.

LM Studio

Download and install the latest version of LM Studio
In the app press cmd/ctrl-shift-m and download at least one model (e.g., qwen2.5-coder-7b). Alternatively, you can get models via the LM Studio CLI:
```
lms get qwen2.5-coder-7b
```
Make sure the LM Studio API server is running by executing:
```
lms server start
```

Tip: Set LM Studio as a login item to automate running the LM Studio server.

Mistral

Visit the Mistral platform and create an API key
Open the configuration view (agent: open configuration) and navigate to the Mistral section
Enter your Mistral API key

The Mistral API key will be saved in your keychain.

Zed will also use the MISTRAL_API_KEY environment variable if it's defined.

Custom Models

The Zed agent comes pre-configured with several Mistral models (codestral-latest, mistral-large-latest, mistral-medium-latest, mistral-small-latest, open-mistral-nemo, and open-codestral-mamba). All the default models support tool use. If you wish to use alternate models or customize their parameters, you can do so by adding the following to your Zed settings.json:

{
  "language_models": {
    "mistral": {
      "api_url": "https://api.mistral.ai/v1",
      "available_models": [
        {
          "name": "mistral-tiny-latest",
          "display_name": "Mistral Tiny",
          "max_tokens": 32000,
          "max_output_tokens": 4096,
          "max_completion_tokens": 1024,
          "supports_tools": true,
          "supports_images": false
        }
      ]
    }
  }
}

Custom models will be listed in the model dropdown in the Agent Panel.

Ollama

Download and install Ollama from ollama.com/download (Linux or macOS) and ensure it's running with ollama --version.

Download one of the available models, for example, for mistral:
```
ollama pull mistral
```
Make sure that the Ollama server is running. You can start it either via running Ollama.app (macOS) or launching:
```
ollama serve
```
In the Agent Panel, select one of the Ollama models using the model dropdown.

Ollama Context Length

Zed has pre-configured maximum context lengths (max_tokens) to match the capabilities of common models. Zed API requests to Ollama include this as the num_ctx parameter, but the default values do not exceed 16384 so users with ~16GB of RAM are able to use most models out of the box.

See get_max_tokens in ollama.rs for a complete set of defaults.

Depending on your hardware or use-case you may wish to limit or increase the context length for a specific model via settings.json:

{
  "language_models": {
    "ollama": {
      "api_url": "http://localhost:11434",
      "available_models": [
        {
          "name": "qwen2.5-coder",
          "display_name": "qwen 2.5 coder 32K",
          "max_tokens": 32768,
          "supports_tools": true,
          "supports_thinking": true,
          "supports_images": true
        }
      ]
    }
  }
}

If you specify a context length that is too large for your hardware, Ollama will log an error. You can watch these logs by running: tail -f ~/.ollama/logs/ollama.log (macOS) or journalctl -u ollama -f (Linux). Depending on the memory available on your machine, you may need to adjust the context length to a smaller value.

You may also optionally specify a value for keep_alive for each available model. This can be an integer (seconds) or alternatively a string duration like "5m", "10m", "1h", "1d", etc. For example, "keep_alive": "120s" will allow the remote server to unload the model (freeing up GPU VRAM) after 120 seconds.

The supports_tools option controls whether the model will use additional tools. If the model is tagged with tools in the Ollama catalog, this option should be supplied, and the built-in profiles Ask and Write can be used. If the model is not tagged with tools in the Ollama catalog, this option can still be supplied with the value true; however, be aware that only the Minimal built-in profile will work.

The supports_thinking option controls whether the model will perform an explicit "thinking" (reasoning) pass before producing its final answer. If the model is tagged with thinking in the Ollama catalog, set this option and you can use it in Zed.

The supports_images option enables the model's vision capabilities, allowing it to process images included in the conversation context. If the model is tagged with vision in the Ollama catalog, set this option and you can use it in Zed.

OpenAI

Visit the OpenAI platform and create an API key
Make sure that your OpenAI account has credits
Open the settings view (agent: open configuration) and go to the OpenAI section
Enter your OpenAI API key

The OpenAI API key will be saved in your keychain.

Zed will also use the OPENAI_API_KEY environment variable if it's defined.

Custom Models

The Zed agent comes pre-configured to use the latest version for common models (GPT-3.5 Turbo, GPT-4, GPT-4 Turbo, GPT-4o, GPT-4o mini). To use alternate models, perhaps a preview release or a dated model release, or if you wish to control the request parameters, you can do so by adding the following to your Zed settings.json:

{
  "language_models": {
    "openai": {
      "available_models": [
        {
          "name": "gpt-4o-2024-08-06",
          "display_name": "GPT 4o Summer 2024",
          "max_tokens": 128000
        },
        {
          "name": "o1-mini",
          "display_name": "o1-mini",
          "max_tokens": 128000,
          "max_completion_tokens": 20000
        }
      ],
      "version": "1"
    }
  }
}

You must provide the model's context window in the max_tokens parameter; this can be found in the OpenAI model documentation.

OpenAI o1 models should set max_completion_tokens as well to avoid incurring high reasoning token costs. Custom models will be listed in the model dropdown in the Agent Panel.

OpenAI API Compatible

Zed supports using OpenAI compatible APIs by specifying a custom endpoint and available_models for the OpenAI provider.

You can add a custom API URL for OpenAI either via the UI or by editing your settings.json. Here are a few model examples you can plug in by using this feature:

X.ai Grok

Example configuration for using X.ai Grok with Zed:

  "language_models": {
    "openai": {
      "api_url": "https://api.x.ai/v1",
      "available_models": [
        {
          "name": "grok-beta",
          "display_name": "X.ai Grok (Beta)",
          "max_tokens": 131072
        }
      ],
      "version": "1"
    },
  }

OpenRouter

OpenRouter provides access to multiple AI models through a single API. It supports tool use for compatible models.

Visit OpenRouter and create an account
Generate an API key from your OpenRouter keys page
Open the settings view (agent: open configuration) and go to the OpenRouter section
Enter your OpenRouter API key

The OpenRouter API key will be saved in your keychain.

Zed will also use the OPENROUTER_API_KEY environment variable if it's defined.

Custom Models

You can add custom models to the OpenRouter provider by adding the following to your Zed settings.json:

{
  "language_models": {
    "open_router": {
      "api_url": "https://openrouter.ai/api/v1",
      "available_models": [
        {
          "name": "google/gemini-2.0-flash-thinking-exp",
          "display_name": "Gemini 2.0 Flash (Thinking)",
          "max_tokens": 200000,
          "max_output_tokens": 8192,
          "supports_tools": true,
          "supports_images": true,
          "mode": {
            "type": "thinking",
            "budget_tokens": 8000
          }
        }
      ]
    }
  }
}

The available configuration options for each model are:

name (required): The model identifier used by OpenRouter
display_name (optional): A human-readable name shown in the UI
max_tokens (required): The model's context window size
max_output_tokens (optional): Maximum tokens the model can generate
max_completion_tokens (optional): Maximum completion tokens
supports_tools (optional): Whether the model supports tool/function calling
supports_images (optional): Whether the model supports image inputs
mode (optional): Special mode configuration for thinking models

You can find available models and their specifications on the OpenRouter models page.

Custom models will be listed in the model dropdown in the Agent Panel.

Vercel v0

Vercel v0 is an expert model for generating full-stack apps, with framework-aware completions optimized for modern stacks like Next.js and Vercel. It supports text and image inputs and provides fast streaming responses.

The v0 models are OpenAI-compatible models, but Vercel is listed as first-class provider in the panel's settings view.

To start using it with Zed, ensure you have first created a v0 API key. Once you have it, paste it directly into the Vercel provider section in the panel's settings view.

You should then find it as v0-1.5-md in the model dropdown in the Agent Panel.

Advanced Configuration

Custom Provider Endpoints

You can use a custom API endpoint for different providers, as long as it's compatible with the provider's API structure. To do so, add the following to your settings.json:

{
  "language_models": {
    "some-provider": {
      "api_url": "http://localhost:11434"
    }
  }
}

Where some-provider can be any of the following values: anthropic, google, ollama, openai.

Default Model

Zed's hosted LLM service sets claude-sonnet-4 as the default model. However, you can change it either via the model dropdown in the Agent Panel's bottom-right corner or by manually editing the default_model object in your settings:

{
  "agent": {
    "version": "2",
    "default_model": {
      "provider": "zed.dev",
      "model": "gpt-4o"
    }
  }
}

Feature-specific Models

If a feature-specific model is not set, it will fall back to using the default model, which is the one you set on the Agent Panel.

You can configure the following feature-specific models:

Thread summary model: Used for generating thread summaries
Inline assistant model: Used for the inline assistant feature
Commit message model: Used for generating Git commit messages

Example configuration:

{
  "agent": {
    "version": "2",
    "default_model": {
      "provider": "zed.dev",
      "model": "claude-sonnet-4"
    },
    "inline_assistant_model": {
      "provider": "anthropic",
      "model": "claude-3-5-sonnet"
    },
    "commit_message_model": {
      "provider": "openai",
      "model": "gpt-4o-mini"
    },
    "thread_summary_model": {
      "provider": "google",
      "model": "gemini-2.0-flash"
    }
  }
}

Alternative Models for Inline Assists

You can configure additional models that will be used to perform inline assists in parallel. When you do this, the inline assist UI will surface controls to cycle between the alternatives generated by each model.

The models you specify here are always used in addition to your default model. For example, the following configuration will generate two outputs for every assist. One with Claude 3.7 Sonnet, and one with GPT-4o.

{
  "agent": {
    "default_model": {
      "provider": "zed.dev",
      "model": "claude-sonnet-4"
    },
    "inline_alternatives": [
      {
        "provider": "zed.dev",
        "model": "gpt-4o"
      }
    ],
    "version": "2"
  }
}

Default View

Use the default_view setting to set change the default view of the Agent Panel. You can choose between thread (the default) and text_thread:

{
  "agent": {
    "default_view": "text_thread"
  }
}